Bivariate Analyses of Revision Risk
We performed bivariate Cox proportional hazards models with time to revision as the dependent variable. Separate models were created for each potential predictor. Younger patient age, male sex, and surgeon volume of fewer than six total hip replacements per year in the Medicare population were all associated with higher revision rates, with hazard ratios for these factors of ≥1.24 (Table II). Hospital volume of fewer than twenty-five total hip replacements per year (compared with more than fifty procedures) had a hazard ratio (HR) of 1.14 (95% confidence interval [95% CI], 1.07, 1.23). All other variables were not significantly associated with revision total hip replacement, with hazard ratios between 0.90 and 1.02, indicating <10% relative change in risk across levels of the variable.
Multivariate Analyses of Revision Risk
Multivariate analyses included all of the independent variables listed in Table I (patient age, sex, race, Medicaid eligibility, and comorbidity and hospital and surgeon volume). The results of these analyses (Table II) showed that younger patient age (sixty-five to seventy-five years [HR, 1.47; 95% CI, 1.37, 1.58) was associated independently with increased revision risk after adjusting for all other variables, as was male sex (HR, 1.23; 95% CI, 1.15, 1.31) and having surgery performed by a low-volume surgeon (fewer than six total hip replacements per year compared with more than twelve per year [HR, 1.21; 95% CI, 1.12, 1.32]). We did not observe a significant interaction between the association of surgeon volume and either patient age or sex on the risk of revision. The hazards for the three strata of surgeon volume differed significantly in the first eighteen months after total hip replacement, with patients of surgeons with the lowest volume having the greatest risk of revision (HR, 1.65; 95% CI, 1.39, 1.97). The hazards did not differ across volume strata over subsequent follow-up.
The competing risk models confirmed the results of the primary analysis, identifying the same principal predictors of revision risk as the primary analysis—younger patient age, male sex, and lower-volume surgeon. Detailed results of the competing risk models are presented in the Appendix.
We examined the risk of revision following a primary total hip replacement in a cohort of 51,347 Medicare beneficiaries who had elective primary total hip replacement between July 1995 and June 1996. Men had higher rates of revision than women, and younger patients (those who were sixty-five to seventy-five years old at the time of primary total hip replacement) had a higher rate than older patients. The risk of dying over the twelve-year follow-up period far exceeded the risk of revision, especially among the cohort of patients over seventy-five years old at the time of primary total hip replacement, for whom the risk of death over the twelve-year follow-up was tenfold greater than the risk of revision.
Our data are consistent with other literature showing an increasing risk of revision associated with younger age and with male sex2,7,19. Also, as we showed previously, the risk of revision was highest in the first eighteen months after surgery, likely reflecting early technical failures10. The influence of surgeon volume on the risk of revision occurs in the first eighteen months postoperatively and not thereafter, suggesting that surgeon experience influences the risk of technical failure but not the risk of failure over the longer term, which likely reflects intrinsic limitations in implant durability. The association of surgeon volume with early failures is similar for younger and older patients, and for men and women. We note that the surgeon volumes in the present report reflect procedures performed in Medicare beneficiaries. Approximately two-thirds of primary total hip replacements are performed in Medicare recipients, and the other third in patients less than sixty-five years old1. Thus, the Medicare volumes reported in this paper can be multiplied by 1.5 to provide a rough estimate of total volumes for the average surgeon. This simplification may not be valid for particular surgeons. Because older patients are at high risk of morbidity and mortality, competing risks may distort estimates of revision risk. We performed analyses that took competing risks into account; these sensitivity analyses largely confirmed the findings of the primary analyses.
The consistent differences in failure rates seen between younger and older patients and between men and women (Fig. 1-A) raise the question of whether shorter follow-up periods could be used to assess differences in implant performance rather than the longer (twelve-year) periods of observation utilized in this study. Such a strategy would be less resource-intensive than twelve-year follow-up. This idea cannot be addressed directly by our study and merits further investigation. However, since surgeon practices may influence early failures, while implant and patient characteristics influence the later ones, this strategy may not capture risk profiles over time accurately.
The study has important limitations. First, ICD-9 coding does not distinguish between right and left-sided procedures. Up to 30% of the revisions that occur following primary total hip replacement are performed on the contralateral side, even with censoring of patients who had a second primary total hip replacement during the follow-up period12. The likelihood that the revision is performed on the index compared with the contralateral hip is not influenced by patient age or sex or surgeon volume, the three factors associated with revision risk in this analysis12. However, the revision risks shown in Figures 1 and 2 may be overstated by as much as 30%, and the misclassification may have blunted the associations we observed. Further, censoring the data on patients who had a contralateral primary total hip replacement over the period of follow-up may have inadvertently undercounted some bona fide revisions of the index hip. The more general point is that administrative data contain relatively little clinical information, and ICD-9 coding does not convey nuanced clinical information perfectly. These issues create important limitations in the inferences that can be derived from studies that are based on administrative data. Thus, our findings must be confirmed in studies that retain the national scope of this analysis but overcome the ambiguities of administrative data.
Second, revision arthroplasty is an ambiguous proxy for total hip replacement failure. Revision requires that surgery be offered to and accepted by the patient. Some patients with symptomatic loosening of the prosthesis may be too sick or frail to undergo surgery, others may simply not present for care, and still others may be offered revision total hip replacement but prefer not to undergo the surgery. Consequently, actual rates of symptomatic failure of total hip replacement are likely to be higher than observed revision rates.
Third, while the length of follow-up is a strength of this study, it also means that the index procedures were done in the mid-1990s. Implant technology, surgical technique, and rehabilitation practice have evolved over the subsequent years. It is conceivable that the associations between covariates and revision documented in these analyses may have been altered because of changes in implant technology and the process of care occurring over the last fifteen years.
The observation that patients over seventy-five years of age undergoing primary total hip replacement face a tenfold higher risk of death than of revision over the subsequent twelve years has important implications for patient decision-making, quality improvement, and research. From a decision-making standpoint, these older patients should place revision risk in perspective as they discuss the advantages and drawbacks of total hip replacement with their physicians. Because they face a 60% risk of death over the next twelve years, their absolute risk of revision is just 6%. Patients in the sixty-five to seventy-five-year-old group face a somewhat higher absolute risk of revision—approximately 10%. By extension, patients less than sixty-five years old (and especially those in their fifties) face still higher absolute risks of revision over a ten to twenty-year time frame.
The observation that patients who have total hip replacement after the age of sixty-five years are much more likely to die than to have a revision total hip replacement points to the importance of optimizing short-term outcomes of total hip replacement in the elderly, such as complications and restoration of function. Long-term survival data on implants, as can potentially be obtained by longitudinal joint replacement registries, will be of particular value in assessing and optimizing the outcomes of younger patients who are much more likely to live long enough to face possible revision surgery.
Further study of the predictors of revision in younger patients would provide a more quantitative assessment of these trade-offs and should be a research priority. These observations also impact the optimal frequency with which to monitor patients with radiographs following surgery. Finally, our findings suggest that the development of innovative technologies to improve implant longevity should target younger populations with advanced arthritis, who have a longer anticipated time span in which the implants might fail and require revision arthroplasty.
Tables showing the incidence and average time for each event type and the results from models adjusting for competing risk are available with the online version of this article as a data supplement at jbjs.org.
Investigation performed at the Orthopedic and Arthritis Center for Outcomes Research, Department of Orthopedic Surgery, Brigham and Women’s Hospital, Boston; the Department of Orthopedic Surgery, Massachusetts General Hospital, Harvard Medical School, Boston, Massachusetts; the Department of Medicine, University of North Carolina, Chapel Hill, North Carolina; and the Department of Orthopedic Surgery, University Health Network, University of Toronto, Toronto, Ontario, Canada
This article was chosen to appear electronically on September 12, 2012, in advance of publication in a regularly scheduled issue.
A commentary by Kevin J. Bozic, MD, MBA, and Steven M. Kurtz, PhD, is linked to the online version of this article at jbjs.org.
1. United States Department of Health & Human Services. HCUP nationwide inpatient sample. 2009.
2. Hallan G Lie SA Furnes O Engesaeter LB Vollset SE Havelin LI. Medium- and long-term performance of 11,516 uncemented primary femoral stems from the Norwegian arthroplasty register. J Bone Joint Surg Br. 2007;89:1574–80.
3. Mäkelä K Eskelinen A Pulkkinen P Paavolainen P Remes V. Cemented total hip replacement for primary osteoarthritis in patients aged 55 years or older: results of the 12 most common cemented implants followed for 25 years in the Finnish Arthroplasty Register. J Bone Joint Surg Br. 2008;90:1562–9.
4. Mäkelä KT Eskelinen A Pulkkinen P Paavolainen P Remes V. Total hip arthroplasty for primary osteoarthritis in patients fifty-five years of age or older. An analysis of the Finnish arthroplasty registry. J Bone Joint Surg Am. 2008;90:2160–70.
5. Espehaug B Furnes O Engesaeter LB Havelin LI. 18 years of results with cemented primary hip prostheses in the Norwegian Arthroplasty Register: concerns about some newer implants. Acta Orthop. 2009;80:402–12.
6. Hailer NP Garellick G Kärrholm J. Uncemented and cemented primary total hip arthroplasty in the Swedish Hip Arthroplasty Register. Acta Orthop. 2010;81:34–41.
7. Johnsen SP Sørensen HT Lucht U Søballe K Overgaard S Pedersen AB. Patient-related predictors of implant failure after primary total hip replacement in the initial, short- and long-terms. A nationwide Danish follow-up study including 36,984 patients. J Bone Joint Surg Br. 2006;88:1303–8.
8. Pedersen AB Svendsson JE Johnsen SP Riis A Overgaard S. Risk factors for revision due to infection after primary total hip arthroplasty. A population-based study of 80,756 primary procedures in the Danish Hip Arthroplasty Registry. Acta Orthop. 2010;81:542–7.
9. Conroy JL Whitehouse SL Graves SE Pratt NL Ryan P Crawford RW. Risk factors for revision for early dislocation in total hip arthroplasty. J Arthroplasty. 2008;23:867–72.
10. Losina E Barrett J Mahomed NN Baron JA Katz JN. Early failures of total hip replacement: effect of surgeon volume. Arthritis Rheum 2004;50:1338–43.
11. Katz JN Losina E Barrett J Phillips CB Mahomed NN Lew RA Guadagnoli E Harris WH Poss R Baron JA. Association between hospital and surgeon procedure volume and outcomes of total hip replacement in the United States medicare population. J Bone Joint Surg Am. 2001;83-A:1622–9.
12. Katz JN Wright EA Baron JA . Predictive value of Medicare claims data for identifying revision of index hip replacement was modest. J Clin Epidemiol. 2011; 64:543–6.
13. Charlson ME Pompei P Ales KL MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis. 1987;40:373–83.
14. Romano PS Roos LL Jollis JG. Adapting a clinical comorbidity index for use with ICD-9-CM administrative data: differing perspectives. J Clin Epidemiol. 1993;46:1075–9, 81-90.
15. Mahomed NN Barrett JA Katz JN Phillips CB Losina E Lew RA Guadagnoli E Harris WH Poss R Baron JA. Rates and outcomes of primary and revision total hip replacement in the United States Medicare population. J Bone Joint Surg Am. 2003;85:27–32.
16. Pepe MS Mori M. Kaplan-Meier, marginal or conditional probability curves in summarizing competing risks failure time data? Stat Med. 1993;12:737–51.
17. Pintilie M. Competing risks: a practical perspective (statistics in practice). West Sussex, England: Wiley; 2006.
18. Fine JP GR. A proportional hazards model for the subdistribution of competing risk. J Am Statist Assoc. 1988;94:496–509.
19. Corbett KL Losina E Nti AA Prokopetz JJ Katz JN. Population-based rates of revision of primary total hip arthroplasty: a systematic review. PLoS One. 2010;5:e13520.
Disclosure: One or more of the authors received payments or services, either directly or indirectly (i.e., via his or her institution), from a third party in support of an aspect of this work. In addition, one or more of the authors, or his or her institution, has had a financial relationship, in the thirty-six months prior to submission of this work, with an entity in the biomedical arena that could be perceived to influence or have the potential to influence what is written in this work. No author has had any other relationships, or has engaged in any other activities, that could be perceived to influence or have the potential to influence what is written in this work. The complete Disclosures of Potential Conflicts of Interest submitted by authors are always provided with the online version of the article.Copyright 2012 by The Journal of Bone and Joint Surgery, Incorporated