The Journal of Arthroplasty Vol. 18 No.

3 2003

Survival Analysis of Primary Cemented Total Knee

Arthroplasty
Which Designs Last?

M. C. Forster, MB ChB, FRCS(Ed)

Abstract: Both posterior stabilization and metal backing of the tibial component are
design changes that could alter the longevity of total knee arthroplasties (TKAs).
Survival analysis data from 16 articles (5,950 knees) were combined to compare
design features. No difference was found in survival between posterior stabilized
implants and implants that were not stabilized or between metal-backed and all-
polyethylene tibial components. Those all-polyethylene tibial components that were
not stabilized showed significantly better survival than metal-backed, nonstabilized
tibial components and posterior stabilized, metal-backed components (P⬍.05) but
not posterior stabilized, all-polyethylene components. According to the currently
available literature, posterior stabilization or metal backing of the tibial component
does not improve the longevity of primary cemented fixed bearing condylar-type
TKAs. Key words: survival analysis, total knee arthroplasty, all-polyethylene,
metal-backed.
© 2003 Elsevier Inc. All rights reserved.

The cemented total condylar knee arthroplasty was
a major advance in total knee arthroplasty (TKA),
which led to good long-term survival of knee im-
plants [1]. Since then, many similar implants have
been developed. Comparison among these different
designs is difficult and requires large numbers be-
cause the failure rate of these implants is low [2]. As
a result, controversy exists over which designs are
best to use.
Posterior stabilization was introduced to improve
stair climbing, increase range of motion, and pre-
vent posterior subluxation of the tibia [3]. Metal
backing of the tibial component was introduced
From the Lincoln County Hospital, Lincoln, United Kingdom.
Submitted November 20, 2001; accepted October 9, 2002.
No benefits or funds were received in support of this study.
Reprint requests: M. C. Forster, MB ChB, FRCS(Ed), 9 Lam-
bourne Avenue, Ashbourne, Derbyshire, UK.
© 2003 Elsevier Inc. All rights reserved.
0883-5403/03/1803-0004$30.00/0
doi:10.1054/arth.2003.50051
after biomechanical studies showed that this pro-
duced better load distribution to the proximal
tibia [4 – 6]. Design changes such as these could
alter the ability of prostheses to last.
In this study, data from previously published
survival analysis life tables of primary total condy-
lar-type TKAs has been combined to enable com-
parison of different design features. The null hy-
pothesis of this study was that there is no difference
between implants in terms of survival. An experi-
mentally tested part of the hypothesis was that
posterior stabilization or metal backing of the tibial
component would significantly improve the lon-
gevity of primary cemented fixed bearing condylar-
type TKAs.
Materials and Methods
MEDLINE (1966 onwards) and EMBASE (1988
onwards) databases were searched to find suitable
references using the following key words: arthro-

265
266 The Journal of Arthroplasty Vol. 18 No. 3 April 2003

Table 1. Survival Analysis Life Table for the Posterior Stabilized Group

95% Confidence
Years Number Failures Withdrawn Number at Risk Failure Rate Success Rate Intervals

0–1 1,698 9 121 1633 0.6% 99.4% 99.0%–99.7%

1–2 1,568 10 348 1389 0.7% 98.7% 98.0%–99.2%
2–3 1,210 5 233 1091 0.5% 98.2% 97.2%–98.8%
3–4 972 5 176 882.5 0.6% 97.6% 96.4%–98.4%
4–5 791 1 130 725.5 0.1% 97.5% 96.1%–98.4%
5–6 660 2 144 587 0.3% 97.2% 95.5%–98.3%
6–7 514 2 54 486 0.4% 96.8% 94.7%–98.1%
7–8 458 4 29 441.5 0.9% 95.9% 93.6%–97.4%
8–9 425 4 44 401 1.0% 94.9% 92.2%–96.7%
9–10 377 3 77 337 0.9% 94.0% 90.6%–96.1%
10–11 296 3 102 243.5 1.2% 92.7% 88.0%–95.4%

plasty, replacement, knee, total knee replacement,
and knee arthroplasty. Any potentially suitable ref-
erence was obtained and examined further. The
bibliographies of these references were also exam-
ined for additional articles that may have been
suitable.
To be included, the article needed to include
results of 5 years or more for primary cemented
fixed bearing condylar-type TKAs, including a sur-
vival analysis table. Series performed on a selected
patient groups (for example younger age, elderly,
or patients with rheumatoid arthritis) were ex-
cluded to reduce possible bias. When 2 series in-
volving the same implant from the same institution
were available, the most recent article with the
longest follow-up was used.
The articles were divided into posterior stabilizing
(PS), nonstabilised, metal-backed (MB), and all-
polyethylene tibial component groups. These were
also further subdivided into 4 groups: posterior
stabilizing–all polyethylene (PSAP), posterior stabi-
lizing–metal-backed (PSMB), nonstabilized–all
polyethylene (NSAP), and nonstabilized–metal-
backed (NSMB) groups. Survival analysis was used
to compare these groups, using revision as the end-
point [2]. The 95% confidence intervals were cal-
culated using Rothman’s method [7].
Results
The MEDLINE and EMBASE searches retrieved
over 3,000 possible articles. From these, 16 suitable
papers were found; these form the basis of this
study [8 –23]. One paper gave results on a posterior
stabilizing TKA with both metal-backed and all-
polyethylene tibial components [15]. This was in-
cluded in the PS group but was not used in any
other groupings. Analysis included 5,950 TKAs. Ta-
bles 1 through 8 show the 10-year survival data for
each group. A ⬎ 90% survival was seen at 10 years
in all groups. No significant difference was seen in
survival between the PS and nonstabilized groups
or between the MB and all-polyethylene groups.
When analysis was performed combining design
features, the NSAP group showed significantly bet-

Table 2. Survival Analysis Life Table for the Nonstabilized Group

95% Confidence
Years Number Failures Withdrawn Number at Risk Failure Rate Success Rate Intervals

0–1 2,218 4 67 2182.5 0.2% 99.8% 99.5%–99.9%

1–2 2,147 9 57 2114 0.4% 99.4% 99.0%–99.6%
2–3 2,081 10 118 2017 0.5% 98.9% 98.3%–99.2%
3–4 1,953 6 90 1905 0.3% 98.6% 98.0%–99.0%
4–5 1,857 7 62 1822.5 0.4% 98.2% 97.5%–98.7%
5–6 1,788 10 208 1679 0.6% 97.6% 96.8%–98.2%
6–7 1,570 9 133 1499 0.6% 97.0% 96.0%–97.8%
7–8 1,428 18 60 1389 1.3% 95.7% 94.5%–96.7%
8–9 1,350 10 440 1125 0.9% 94.8% 93.3%–96.0%
9–10 900 8 217 787.5 1.0% 93.8% 91.9%–95.3%
10–11 773 10 56 740 1.4% 92.4% 90.3%–94.1%
 Which Total Knee Arthroplasties Last? • M. C. Forster 267

Table 3. Survival Analysis Life Table for the Metal-Backed Tibial Component Group

95% Confidence
Years Number Failures Withdrawn Number at risk Failure rate Success rate intervals

0–1 1,561 9 113 1500 0.6% 99.4% 98.9%–99.7%

1–2 1,439 10 338 1265 0.8% 98.6% 97.8%–99.1%
2–3 1,091 3 197 991 0.3% 98.3% 97.3%–98.9%
3–4 891 3 157 811 0.4% 97.9% 96.7%–98.7%
4–5 731 1 110 675.5 0.1% 97.8% 96.3%–98.4%
5–6 620 1 114 562.5 0.2% 97.6% 96.0%–98.4%
6–7 505 1 50 479.5 0.2% 97.4% 95.6%–98.5%
7–8 454 3 26 439.5 0.7% 96.7% 94.6%–98.0%
8–9 425 4 44 401 1.0% 95.7% 93.2%–97.3%
9–10 377 4 77 336.5 1.2% 94.5% 91.5%–96.5%
10–11 296 4 138 225 1.8% 92.7% 89.4%–94.6%

ter survival than the NSMB and PSMB groups
(P⬍.05) but not the PSAP group.
Discussion
It is important that, once implanted, knee pros-
theses last. By collating results from many series,
we hope to minimize the influence of patient and
surgeon factors and gain some insight into implant
factors. However, any meta-analysis is subject to an
element of publication bias. Only published work is
included in the study, but only if these studies
represent all TKAs will the meta-analysis be valid.
Excluding studies on selected patient groups mini-
mizes any bias and makes this meta-analysis as
reliable as possible.
Meta-analysis enables the collection of a large
amount of data (5,950 knees in this case). Random-
ized controlled clinical trials would of course pro-
vide better information but would be difficult to run
on the same scale. These have not yet been per-
formed. This method of analysis allows some insight
into implant survival using the currently available
data.
There are many other papers describing TKAs but
written in a multiple case report style without sur-
vival analysis. Without a life table, we could per-
form survival analysis; therefore, these studies
could not be included. The inclusion of a survival
analysis life table in all series of arthroplasties
would facilitate many other meta-analyses, and
would have made this analysis considerably larger.
Both posterior stabilization and metal backing of
the tibial component are changes that could alter
longevity. The increased conformity of the posterior
stabilized insert should lead to lower contact
stresses and reduced wear. Conversely, horizontal
forces between the cam and socket could lead to
increased forces at the cement– bone interface and
potentially increased aseptic loosening [24]. The
stabilization peg articulation is also an additional

Table 4. Survival Analysis Life Table for the All-Polyethylene Tibial Component Group

95% Confidence
Years Number Failures Withdrawn Number at Risk Failure Rate Success Rate Intervals

0–1 473 1 6 469.5 0.2% 99.8% 98.8%–100%

1–2 466 3 9 460 0.7% 99.2% 98.0%–99.8%
2–3 454 2 11 447.5 0.4% 98.7% 97.2%–99.4%
3–4 441 1 9 436 0.2% 98.5% 96.9%–99.3%
4–5 431 2 8 426 0.5% 98.0% 96.2%–99.0%
5–6 421 0 14 414 0.0% 98.0% 96.1%–98.5%
6–7 407 0 19 397.5 0.0% 98.0% 96.1%–99.0%
7–8 388 1 15 380.5 0.3% 97.7% 95.6%–98.8%
8–9 372 3 17 362 0.8% 96.9% 94.5%–98.3%
9–10 352 2 42 330 0.6% 96.3% 93.7%–97.9%
10–11 308 1 96 259.5 0.4% 95.9% 92.7%–97.8%
268 The Journal of Arthroplasty Vol. 18 No. 3 April 2003

Table 5. Survival Analysis Life Table for the Nonstabilized, Metal-Backed Tibial Component Group

95% Confidence
Year Number Failures Withdrawn Number at Risk Failure Rate Success Rate Intervals

0–1 2034 4 57 2003.5 0.2% 99.8% 99.5%–99.9%

1–2 1973 9 67 1935 0.5% 99.3% 98.8%–99.6%
2–3 1897 10 118 1833 0.5% 98.8% 98.4%–99.3%
3–4 1769 6 90 1721 0.3% 98.5% 97.8%–99.0%
4–5 1673 6 59 1640.5 0.4% 98.1% 97.4%–98.7%
5–6 1608 10 204 1501 0.7% 97.4% 96.5%–98.1%
6–7 1394 9 128 1325.5 0.7% 96.7% 95.6%–97.5%
7–8 1257 18 165 1165.5 1.5% 95.2% 93.8%–96.3%
8–9 1174 9 429 955 0.9% 94.3% 92.6%–95.6%
9–10 736 8 109 677.5 1.2% 93.1% 90.9%–94.8%
10–11 619 10 50 589 1.7% 91.4% 88.8%–93.4%

Table 6. Survival Analysis Life Table for the Posterior Stabilized, Metal-Backed Tibial Component Group

95% Confidence
Years Number Failures Withdrawn Number at Risk Failure Rate Success Rate Intervals

0–1 1272 8 107 1214.5 0.7% 99.3% 98.7%–99.6%

1–2 1157 7 329 989 0.7% 98.6% 97.7%–99.2%
2–3 821 1 186 727.5 0.1% 98.5% 97.3%–99.2%
3–4 634 2 148 559 0.4% 98.1% 96.6%–99.0%
4–5 484 0 105 431.5 0.0% 98.1% 96.3%–99.0%
5–6 379 1 104 326.5 0.3% 97.8% 95.6%–98.9%
6–7 274 1 36 255.5 0.4% 97.4% 94.6%–98.8%
7–8 237 2 12 230 0.9% 96.5% 93.2%–98.3%
8–9 223 2 19 212.5 0.9% 95.6% 91.7%–97.7%
9–10 189 2 45 165.5 1.2% 94.4% 89.6%–97.0%
10–11 142 3 48 116.5 2.6% 92.8% 83.7%–94.2%

Table 7. Survival Analysis Life Table for the Posterior Stabilized, All-Polyethylene Tibial Component Group

95% Confidence
Years Number Failures Withdrawn Number at Risk Failure Rate Success Rate Intervals

0–1 289 1 6 285.5 0.4% 99.6% 98.0%–100%

1–2 282 3 9 276 1.1% 98.5% 96.3%–99.5%
2–3 270 2 11 263.5 0.8% 97.7% 95.1%–98.9%
3–4 257 1 9 252 0.4% 97.3% 94.5%–98.7%
4–5 247 1 5 243 0.4% 96.9% 93.9%–98.5%
5–6 241 0 10 236 0.0% 96.9% 93.8%–98.6%
6–7 231 0 14 224 0.0% 96.9% 93.7%–98.5%
7–8 217 1 14 209.5 0.5% 96.4% 92.9%–98.3%
8–9 202 2 12 195 1.0% 95.4% 91.5%–97.7%
9–10 188 2 32 171 1.2% 94.2% 89.6%–97.0%
10–11 154 1 90 108.5 0.9% 93.3% 86.9%–96.9%

Table 8. Survival Analysis Life Table for the Nonstabilized, All-Polyethylene Tibial Component Group.

95% Confidence
Years Number Failures Withdrawn Number at Risk Failure Rate Success Rate Intervals

0–1 184 0 0 184 0.0% 100% 97.9%–100%

1–2 184 0 0 184 0.0% 100% 97.9%–100%
2–3 184 0 0 184 0.0% 100% 97.9%–100%
3–4 184 0 0 184 0.0% 100% 97.9%–100%
4–5 184 1 3 182 0.5% 99.5% 97.0%–100%
5–6 180 0 4 178 0.0% 99.5% 97.0%–100%
6–7 176 0 5 174.5 0.0% 99.5% 96.9%–100%
7–8 171 0 1 170.5 0.0% 99.5% 97.0%–100%
8–9 170 1 5 167 0.6% 98.9% 95.9%–99.8%
9–10 164 0 10 159 0.0% 98.9% 95.8%–99.8%
10–11 154 0 6 151 0.0% 98.9% 95.7%–99.7%
 Which Total Knee Arthroplasties Last? • M. C. Forster
source of wear debris [25]. These factors may ex-
plain the lack of benefit from posterior stabilization
seen in our study.
No prospective randomized trials comparing
TKAs with and without posterior stabilization were
noted. Evidence from retrospective and prospective
cohort studies suggest no difference in survival for
up to 5 years [26,27]. No comparative studies with
a 10-year follow-up period were noted.
Metal backing of the tibial component gives ex-
perimentally better load distribution to the proxi-
mal tibia [4 – 6]. Theoretically, this improvement
in load distribution should reduce aseptic loosening;
however, this has not been shown in practice
[28 –31]. To our knowledge, only one randomized
prospective trial comparing all-polyethylene and
metal-backed tibial components has been pub-
lished [28]. In this study, 324 TKAs were performed
in 296 patients over 60 years of age, and the results
of 213 joints (66%) were reported at a minimum
follow-up period of 3 years. No difference was seen
in terms of survival at that stage, but the length of
follow-up is short. Evidence from retrospective
studies suggests no difference in survival between
all-polyethylene and metal-backed tibial compo-
nents for up to 10 years [28 –31]. However, these
studies have small numbers, short follow-up times,
or both.
The extra thickness of the metal base is obtained
either at the expense of greater tibial resection or
thinner polyethylene. Greater tibial resection may
lead to implanting onto weaker metaphyseal bone,
and thin polyethylene is more vulnerable to wear,
particularly delamination [32]. These factors may
explain the lack of benefit from metal backing seen
in this study.
Metal backing has led to the increasing modular-
ity of implants. There are advantages in having a
single knee arthroplasty system that can accommo-
date a wide range of knee problems, from straight-
forward primary knee arthroplasty to the challeng-
ing revision. With a metal base plate, stems, blocks,
or wedges may be added, but these components are
rarely required for primary TKA. Metal backing
does allow exchange of inserts, if this is necessary.
However, knee implants rarely fail in this manner.
Even when they do, revision of the insert alone
should be approached with caution because it may
lead to early failure [33].
Using the currently available literature, this study
supports the view that neither metal backing of the
tibial component nor the use of a posterior stabiliz-
ing insert improves the longevity of primary ce-
mented fixed bearing condylar-type TKAs.
269
References
1. Ranawat CS, Boachie-Adjei O: Survivorship analysis
and results of total condylar knee arthroplasty: eight-
to 11- year follow-up period. Clin Orthop 226:6,1988
2. Murray DW, Carr AJ, Bulstrode C: Survival analysis
of joint replacements. J Bone Joint Surg Br 75:697,
1993
3. Insall JN, Lachiewicz PF, Burstein AH: The posterior
stabilised condylar prosthesis: a modification of the
total condylar design. Two- to four-year clinical ex-
perience. J Bone Joint Surg Am 64:1317, 1982
4. Bartel DL, Burstein AH, Santavicca EA, Insall JN:
Performance of the tibial component in total knee
replacement. J Bone Joint Surg Am 64:1026, 1982
5. Lewis JL, Askew MJ, Jaycox DP: A comparative eval-
uation of tibial component designs of total knee pros-
theses. J Bone Joint Surg Am 64:129, 1982
6. Reilly D, Walker PS, Ben-Dov M, Ewald FC: Effects of
tibial components on load transfer in the upper tibia.
Clin Orthop 165:273, 1982
7. Rothman KJ: Estimation of confidence limits for the
cumulative probability of survival in life table analy-
sis. J Chronic Dis 31:557, 1978
8. Gill GS, Joshi AB, Mills DM: Total condylar knee
arthroplasty: 16- to 21- year results. Clin Orthop
367:210, 1999
9. Stern SH, Insall JN: Posterior stabilized prosthesis:
results after follow-up of nine to twelve years. J Bone
Joint Surg Am 74:980, 1992
10. Thadani PJ, Vince KG, Ortaaslan SG, et al: Ten- to
12-year follow-up of the Insall-Burstein I total knee
prosthesis. Clin Orthop 380:17, 2000
11. Ranawat CS, Flynn WF, Saddler S, et al: Long-term
results of the total condylar knee arthroplasty: a
15-year survivorship study. Clin Orthop 286:94,
1993
12. Scuderi GR, Insall JN, Windsor RE, Moran MC: Sur-
vivorship of cemented knee replacements. J Bone
Joint Surg Br 71:798, 1989
13. Li PLS, Zamora J, Bentley G: The results at ten years
of the Insall-Burstein II total knee replacement: clin-
ical, radiological and survivorship studies. J Bone
Joint Surg Br 81:647, 1999
14. Emmerson KP, Moran CG, Pinder IM: Survivorship
analysis of the kinematic stabilizer total knee replace-
ment: a 10- to 14-year follow-up. J Bone Joint Surg
Br 78:441, 1996
15. Scott WN, Rubinstein M, Scuderi G: Results after
knee replacement with a posterior cruciate-substitut-
ing prosthesis. J Bone Joint Surg Am 70:1163, 1988
16. van Loon CJM, Wisse MA, de Waal Malefijt MC, et
al: The kinematic total knee arthroplasty: a 10- to
15-year follow-up and survival analysis. Arch Orthop
Trauma Surg 120:48, 2000
17. Ansari S, Ackroyd CE, Newman JH: Kinematic pos-
terior cruciate ligament retaining total knee replace-
ments: a 10-year survivorship study of 445 arthro-
plasties. Am J Knee Surg 11:9, 1998
270 The Journal of Arthroplasty Vol. 18 No. 3 April 2003
18. Weir DJ, Moran CG, Pinder IM: Kinematic condylar
total knee arthroplasty: 14-year survivorship analysis
of 208 consecutive cases. J Bone Joint Surg Br 78:
907, 1996
19. Schai PA, Thornhill TS, Scott RD: Total knee arthro-
plasty with the PFC system: results at a minimum of
ten years and survivorship analysis. J Bone Joint
Surg Br 80:850, 1998
20. Back DL, Cannon SR, Hilton A, et al: The kinemax
total knee arthroplasty: nine years’ experience.
J Bone Joint Surg Br 83:359, 2001
21. Calder JDF, Ashwood N, Hollingdale JP: Survivorship
analysis of the “Performance” total knee replace-
ment: 7-year follow-up. Int Orthop 23:100, 1999
22. Robertsson O, Scott G, Freeman MAR: Ten-year sur-
vival of the cemented Freeman-Samuelson primary
knee arthroplasty. J Bone Joint Surg Br 82:506, 2000
23. Gill GS, Joshi AB: Long-term results of kinematic
condylar knee replacement: an analysis of 404 knees.
J Bone Joint Surg Br 83:355, 2001
24. Insall JN: Historic development, classification and
characteristics of knee prostheses. P. 691. In Insall
JN, Windsor RE, Scott WN, et al. (eds): Surgery of the
knee, ed 2. Churchill Livingstone, New York, 1993
25. Puloski SKT, McCalden RW, MacDonald SJ, et al:
Tibial post wear in posterior stabilised total knee
arthroplasty. J Bone Joint Surg Am 83:390, 2001
26. Laskin RS: Cemented total knee replacement in pa-
tients with osteoarthritis: a five-year follow-up study
using a prosthesis allowing both retention and resec-
tion of the posterior cruicate ligament. Knee 4:1,
1997
27. Pereira DS, Jaffe FF, Ortiguera C: Posterior cruciate
ligament-sparing versus posterior cruicate ligament-
sacrificing arthroplasty: functional results using the
same prosthesis. J Arthroplasty 13:138, 1998
28. Gioe TJ, Bowman KR: A randomised comparison of
all-polyethylene and metal-backed tibial compo-
nents. Clin Orthop 380:108, 2000
29. Rand JA: Comparison of metal-backed and all-poly-
ethylene tibial components in cruciate condylar total
knee arthroplasty. J Arthroplasty 8:307, 1993
30. Apel DM, Tozzi JM, Dorr LD: Clinical comparison of
all-polyethylene and metal-backed tibial components
in total knee arthroplasty. Clin Orthop 273:243, 1991
31. Udomkiat P, Dorr LD, Long W: Matched-pair analysis
of all-polyethylene versus metal-backed tibial com-
ponents. J Arthroplasty 16:689, 2001
32. Insall JN, Haas SB: Complications of total knee ar-
throplasty. P. 912. In Insall JN, Windsor RE, Scott
WN, et al. (eds): Surgery of the knee. Ed 2. Churchill
Livingstone, New York, 1993
33. Babis GC, Trousdale RT, Morrey BF: The effective-
ness of isolated tibial insert exchange in revision total
knee arthroplasty. J Bone Joint Surg Am 84:64, 2002