Undoubtedly, the success of total hip arthroplasty and its worldwide acceptance are due to the durable cemented low-friction arthroplasty devised and popularized by Sir John Charnley1. Although initial acetabular components were cemented, cementless acetabular components have gained popularity over the years and have become the primary components of choice in North America. This is despite the availability of ample literature supporting the use of cemented acetabular components2-6 and suggesting a less optimal survivorship for cementless components7-14. However, there is a lack of long-term studies on the improved cementless implants that may have the potential for better survivorship and durability.
Cementless modular acetabular prostheses were introduced in the 1980s as a reaction to the perceived poor results of cemented all-polyethylene acetabular implants. The goal was to eliminate cement as a fixation modality, as bone cement was thought to be the principal cause of pelvic osteolysis. The latter hypothesis was refuted when long-term studies demonstrated that extensive osteolysis could also occur with cementless components and that greater wear and more extensive osteolysis were observed with cementless porous-coated cups15,16. It has subsequently become apparent that polyethylene wear and expansion of the effective joint space by hydrostatic fluid flow, rather than “cement disease,” are the major causes of pelvic osteolysis9,17.
Both cemented and cementless acetabular components are heterogeneous, with a range of specifications that can affect the revision rate and survivorship. Moreover, various factors such as the material shape and geometry, surface finish, and bearing surface affect the durability of a component, and the surgical approach used and the expertise of the surgeon may also affect its survivorship18. Other important factors that can influence the in vivo wear properties of the bearing surface are the shelf life and sterilization environment (air or an inert environment) of the polyethylene19,20. Regardless of these factors, the method of fixation of the acetabular component also affects survivorship of this component.
The purpose of this study was to examine published clinical results to assess the long-term outcomes of acetabular components in primary total hip arthroplasty in order to compare the outcomes of cemented and cementless acetabular components as measured by the survivorship or revision rate.
Search Strategy and Selection Criteria
A systematic literature search and systematic review was performed with use of PubMed to identify relevant studies up to June 2011 containing the following keywords: “(total) hip replacement,” “hip arthroplasty,” “cemented,” “cementless,” or “uncemented.” No limitation was employed regarding the underlying etiology of the joint disorder that necessitated the hip replacement. Inclusion and exclusion criteria for determining whether an identified article would be eligible for the review were established prior to initiating the search. Inclusion criteria were (1) English language, (2) a retrospective or prospective study of primary total hip arthroplasty with a minimum of ten years of follow-up, and (3) separate reporting on the survivorship of the acetabular component, with revision arthroplasty for any reason as the end point in the corresponding cohort, or reporting on the rate of revision for the acetabular component for any reason.
Exclusion criteria were (1) inclusion of revision total hip arthroplasty cases, (2) a nonclinical study (e.g., a study performed on animals or in vitro, or a mechanical or biomechanical study on cadavers), (3) a general description of a surgical technique or a clinical scoring system (e.g., the Harris hip score) without reporting of the survivorship or revision rate of the acetabular component, (4) reporting of outcomes that involved only the results of radiographic follow-up, and (5) reporting of the survivorship or revision rate of only the femoral component. When a published updated study involving the same cohort of patients was identified, only the latest update was included in the analysis.
Articles that met the inclusion criteria and did not meet the exclusion criteria were selected to be reviewed. The reference lists of the selected articles were also reviewed manually to identify any articles not found in the primary search (Fig. 1).
Outcome Measures
The long-term survivorship and/or revision rate of the acetabular component for any reason at more than ten years postoperatively represented the primary outcome measure in this systematic review. The revision rate and/or survivorship of the acetabular component for aseptic loosening represented a secondary end point.
Data Extraction
Two observers independently reviewed the titles and abstracts of the identified articles. If both observers agreed that a study did not meet the eligibility criteria, it was excluded. After screening of titles and abstracts, the full text of each of the remaining 206 articles was obtained and reviewed by the same two observers independently. Any disagreements were resolved by means of discussion with other members of the reviewing team. Sixty-four articles from the PubMed search were deemed eligible, and seventeen additional articles were added following manual searching of the reference lists of the former articles.
Data on the patient demographics, number of hips treated with total hip arthroplasty, underlying etiology leading to the total hip arthroplasty, duration of follow-up, number of hips or patients lost to follow-up, type of fixation of the acetabular component, the type of implant used, and survivorship or revision rate of the acetabular component were extracted and entered in a spreadsheet. If an article contained multiple subgroups of patients, only the data for the subgroup meeting the inclusion criteria were entered.
Statistical Methods
Two outcomes, the revision rate and survivorship, were compared across the studies. The number of arthroplasties and the total number with the outcome of interest were extracted from each study. Two types of statistical analyses were performed on the extracted data. The first involved meta-analysis of the combined data from the seven articles that directly compared cemented and cementless acetabular components. The meta-analysis was performed with use of Review Manager (version 5.1 [2011]; The Nordic Cochrane Centre, The Cochrane Collaboration, Copenhagen, Denmark). Because of the known variations among the studies and the resulting heterogeneity in the effect sizes (significant at p = 0.005 for survival and borderline at p = 0.08 for revision), a random-effects Mantel-Haenszel model was used.
The second statistical comparison involved study-level logistic regression on the pooled data across all eighty-one relevant studies. The regression was carried out with use of the “glm” function in the R software package (version 2.14.1 [2011]; R Development Core Team, Vienna, Austria). The logistic regression models included overdispersion, which was expected because of the high variation among the studies. Weighting on the basis of the number of total hip arthroplasties was performed in the logistic regression models. The regression was modeled as a function of (n, N − n) with use of the quasibinomial family to account for the high variation in the pooled binary data among the studies. Three types of models were used: (1) a simple model that considered only the type of acetabular component fixation as a predictor and was most directly comparable with the meta-analysis results; (2) an age-adjusted model that considered the fixation type, whether the mean age was at least fifty-five years, and the interaction of fixation type with age; and (3) a full model that expanded on the age-adjusted model by adding the percentage of female patients, the preoperative diagnosis, the mean duration of follow-up, whether the study was prospective or retrospective, and the interaction of age with all of these terms. We did not adjust for the year of publication because many recent publications were actually updated reports of previous cohorts at longer follow-up, and adjusting for the year of publication would therefore not have helped to track the secular trends in practice over time.
Source of Funding
No external funding was received in support of this work.
The studies included 26,576 primary total hip arthroplasties; cemented acetabular fixation was used in 13,509 and not used in 13,067. The preoperative diagnoses (underlying disease leading to degeneration of the affected hip), demographics, types of implants used, and clinical results extracted from each of the included studies are summarized in the Appendix.
When data from all of the articles were pooled and logistic regression was performed with use of a quasibinomial family, the estimated odds ratio (OR) for survival of a cemented acetabular component compared with a cementless component was 1.57 (95% confidence interval [CI], 1.15 to 2.14; p = 0.005). The estimated OR for revision of a cemented cup for any reason compared with revision of a cementless cup was 0.54 (95% CI, 0.30 to 0.98; p = 0.05). When the results were adjusted for age and other demographic variables, there was no significant difference between the rate of revision of cemented and cementless acetabular components (p = 0.46). However, cemented components continued to have significantly greater survivorship than cementless cups; the estimated OR for survival for a cemented compared with a cementless acetabular component was 1.60 (95% CI, 1.32 to 2.40; p = 0.02).
When the end point of revision for aseptic loosening was analyzed, no significant difference between the cemented and cementless cups was found. The OR for revision of a cemented cup for aseptic loosening was 1.06 (95% CI, 0.28 to 4.05; p = 0.92), and the OR for survivorship was 2.4 (95% CI, 0.28 to 4.05; p = 0.19).
In another secondary analysis, the pooled cohort was divided into a younger age group with a mean age of less than fifty-five years and an older age group with a mean age of fifty-five years or more. The survivorship (and standard error) of a cemented cup, with revision for any reason as the end point, was 81.01% ± 1.35% in the younger age group and 88.03% ± 0.34% in the older age group. The survivorship of a cementless cup was 80.95% ± 0.72% in the younger age group and 82.62% ± 0.39% in the older age group. The rate of revision of a cemented cup for any reason was 28.23% ± 2.20% in the younger age group and 7.73% ± 0.31 in the older age group. The rate of revision of a cementless cup for any reason was 17.67% ± 1.25% in the younger age group and 8.91% ± 0.69% in the older age group. This set of values represents an example of Simpson’s paradox: although the mean values for the cementless group were worse than those for the cemented group, younger patients with a cemented cup did worse than such patients with uncemented cups but older patients with cemented cups did better than such patients with uncemented cups.
Seven studies compared either the survivorship or the revision rate of a cemented acetabular component with that of a cementless component (see Appendix). The survivorship and the revision rate of the acetabular component were each reported in five of these studies.
The meta-analysis indicated an OR of 1.49 (95% CI, 0.7 to 3.17) for survival of a cemented acetabular component compared with a cementless component. The OR for revision of a cemented acetabular component compared with a cementless component was 0.7 (95% CI, 0.39 to 1.25). The estimated OR values suggested a benefit (higher survivorship and lower revision rate) from cement fixation, but the effect was not significant (p = 0.30 for survivorship and p = 0.23 for revision). Figures 2 and 3 depict forest plots for survivorship and revision rate in the studies comparing cemented with cementless acetabular components.
There is no dispute that total hip arthroplasty is an effective and successful procedure, and its utilization is expected to increase over the coming years21. It is estimated that approximately 200,000 primary total hip arthroplasties and 36,000 revision total hip arthroplasties were performed in 2003 in the United States22 and that the number of primary and revision total hip arthroplasties will increase by 137% between 2005 and 203021. As failure of the acetabular component accounts for a large proportion of the revisions, it is reasonable to seek strategies that will minimize the need for revision arthroplasty. One of the factors affecting the durability of the acetabular component involves the type of fixation of the component to the pelvis. Currently, the majority of total hip arthroplasties performed in North America utilize an uncemented acetabular component. The usage of uncemented acetabular components during primary total hip arthroplasty is also on the rise in other countries. We are not aware of any literature that supports the superiority of an uncemented acetabular component over a cemented one18,23.
The current controversy regarding the optimal choice of acetabular fixation for total hip arthroplasty clearly originates from the various outcomes reported to date. The heterogeneities in the patient cohorts, materials and bearing surfaces utilized, and study designs each introduce a multitude of variables that skew the reported performance of cemented and cementless acetabular components. Thus, there is a clear necessity for an investigation performed across studies to determine if either acetabular fixation method (cemented or cementless) in total hip arthroplasty is linked to superior performance.
Previous systematic reviews18,23 pooled studies with short-term and long-term outcomes, which may affect the conclusion drawn from these studies. The meta-analysis by Morshed et al. included twenty studies comparing total hip arthroplasty performed with and without cement, but some of these studies had a follow-up duration as short as one year18. Obviously, defining the survivorship of an implant on the basis of such a short-term follow-up could be misleading as inclusion of such results may partially mask survivorship differences in a meta-analysis. The systematic review by Yahiro et al.23 used similar follow-up criteria and therefore also included studies with short and intermediate-term follow-up. Both meta-analyses failed to show better survival of cementless cups compared with their cemented counterparts.
In some studies, failure of an implant was defined as revision for aseptic loosening23. However, in our view, a survival analysis with revision for any cause as the end point more adequately reflects the outcome of interest. Furthermore, the results of a survival analysis with revision as the end point are more predictable than those of an analysis based only on aseptic loosening, as the results of the latter analysis will depend on the definition of aseptic loosening that is selected24. Nonrevised hips are not necessarily successful, and many cup failures are symptomless for long periods, with loss of bone stock occurring in the absence of pain and disability25. Nevertheless, use of revision as an end point is a more objective and reliable approach for assessing the outcome of total hip arthroplasty. We also believe that a liner exchange in a cementless implant should be considered as a failure of the acetabular component, although the complexity and difficulty of a liner exchange procedure are often less than those of a cup revision for aseptic loosening. The authors of some studies have reported greater survival of cementless compared with cemented cups when the end point was revision for aseptic loosening but no difference in survival when the end point was revision for any reason (including liner exchange)11. The present meta-analysis revealed no significant difference in survival between cementless and cemented acetabular components when the end point was revision for aseptic loosening. However, the survival of cemented acetabular components appeared to be better than that of cementless components when the end point included failure and revision for all causes.
The present study has some limitations. A few Level-I studies26-30 on this topic have been published, but the majority of the evaluated studies had a level of evidence of II or III, which reduced the strength of the evidence. Moreover, only seven of the included studies compared the long-term outcomes of cemented and cementless acetabular components, and the availability of matched data (to control for effects of important confounders that can influence survivorship of the acetabular component) was relatively limited. The remaining studies included reports on either cemented or cementless components but did not compare the outcome of these two types.
It is known that the outcome of total hip arthroplasty in general, and the survivorship of the acetabular component in particular, can be influenced by many variables other than the type of fixation, such as the polyethylene shelf life and method of sterilization, patient activity level, bone quality, and body mass index. Prior studies have not always adjusted for these important variables. Indeed, in the majority of the included studies, these variables were either not controlled for or not consistently reported. Consequently, given the large number of factors that can affect survival of an acetabular cup in total hip arthroplasty and the wide heterogeneity among these studies, we were not able to control for these variables in the meta-analysis with any degree of confidence.
Another limitation of the present study is related to the survival analyses in the included studies. Most of the included studies used the Kaplan-Meier method to estimate the probability of implant survival. Since the rate of death increases during long-term follow-up and death is a competing event in the Kaplan-Meier estimator, use of this method in the studies with a high rate of death (e.g., the studies by Callaghan et al.6 and Berry et al.5) is inappropriate and leads to either overestimation of the revision risk or underestimation of the survival probability31. In an attempt to resolve this problem, Murray et al.32 have suggested using a worst-case scenario analysis in which patients lost to follow-up are considered as failures. However, worst-case scenario analysis is not recommended for two reasons: it requires extrapolation of an unknown outcome, and it implies that censoring is not independent of the event mechanism, a situation in which the Kaplan-Meier method would not be valid31.
Another confounding factor is related to the observed improvement in survivorship over time of successive generations of cementless acetabular components. Some studies have indicated no difference in the survival rate between cemented and cementless cups in long-term follow-up when newer-generation cementless cups were used, whereas the older-generation cementless cups had less favorable results than cemented cups11,18. It is certain that earlier designs of cementless acetabular components were responsible for substantial underperformance of these components. For example, threaded acetabular components with a smooth surface treatment had the worst survival rate in the long-term studies8,14. However, inclusion of these studies in the present meta-analysis will not impact the results and comparisons substantially because many studies involving early cemented acetabular components with crude cementing techniques and poor performance were also included33,34.
Despite the aforementioned limitations, the findings of this study are important because they highlight the lack of sufficient evidence for superior survivorship of cementless acetabular components at a minimum of ten years of follow-up. Ongoing critical review to twenty years and beyond is required because the advantage of one fixation method over another may become apparent at long-term follow-up. Future studies can also take into account the role that alternative bearing surfaces may play in enhancing survivorship of cementless acetabular components.