Total knee replacement is a highly effective treatment for end-stage knee arthritis. The utilization of total knee replacement increased by >10% per year in the late 1980s1 and continues to climb, with >350,000 primary total knee replacement procedures and 29,000 revision procedures performed in the United States in 20022. Although total knee replacement is one of the most frequently performed orthopaedic procedures in the United States, there have been few contemporary population-based studies on the utilization of total knee replacement and the variation in utilization according to geographic region and patient gender, race, and socioeconomic status3-8. The population-based perspective is critically important because referral center data do not capture national trends or regional, racial, and other differences in utilization.
Data regarding the complications of total knee replacement are similarly limited. Most of the literature on the outcomes of total knee replacement has been reported from referral centers9, where a minority of these procedures are performed6 and where outcomes may be affected by referral bias and therefore may not be representative. The present study aims to address these issues by reporting on the utilization and short-term outcomes of primary and revision total knee replacement in the United States Medicare population, which comprises virtually all United States residents who are sixty-five years old or more.
Materials and Methods
The present study involves the use of Medicare claims submitted by hospitals (Medicare Part A) or by surgeons (Medicare Part B) in the year 2000. Patients undergoing primary total knee replacement were identified on the basis of a hospital claim with an ICD-9-CM (International Classification of Diseases, Ninth Revision, Clinical Modification) code of 81.54 or a surgeon's claim with a CPT (Current Procedural Terminology) code of 27447. Patients undergoing revision total knee replacement were identified on the basis of a surgeon's claim with a CPT code of 27486 or 27487. Because occasional miscodes in Medicare data occur, a single surgeon's claim was accepted only if the patient was in the hospital on the date of the claim. To focus on elective total knee replacement, we excluded patients with claims indicating preexisting infection of the knee, metastatic cancer, or bone cancer from both study cohorts. In order to obtain complete claims histories, we also excluded patients enrolled in health maintenance organizations and those not enrolled in both parts of Medicare, patients who were less than sixty-five years of age, and patients who were not residents of the United States. On the basis of these criteria, 12,373 (8%) of patients in the primary total knee replacement cohort and 1135 (9%) of those in the revision cohort were excluded. In addition, patients undergoing bilateral primary total knee replacement during the same hospitalization were also excluded; this group comprised an additional 9243 (6%) of the patients in the original primary total knee replacement cohort.
To compute age and gender-specific incidence rates for unilateral primary and revision total knee replacement, we divided the number of procedures observed in a given age-gender demographic group during the study year by an estimate of the person-years at risk in the corresponding group of eligible Medicare enrollees. To do this, we used population data from the year 2000 Medicare denominator file for the first and last dates of our study period (January 1 and December 31, 2000) and counted the number of enrollees who were alive, who were residents of the United States, who were sixty-five years old or more, who were enrolled in both parts of Medicare, and who were not enrolled in a health maintenance organization. For each age-gender demographic group, we averaged the numbers of eligible enrollees at the beginning and end of the year to obtain denominators for annual rates. Identical methods were used to compute incidence rates specific to each Census division of the United States.
Rate ratios, estimated with use of Poisson regression, were used to assess the relationship between demographic characteristics and incidence rates of unilateral primary and revision total knee replacement. The patient characteristics examined in the Poisson regressions were age (coded in five-year age-groups from sixty-five to eighty-nine years of age, with ages between ninety and ninety-nine years combined), gender, race (only enrollees coded as black or white were retained for these analyses because other race/ethnicity categories were not coded reliably), whether or not the enrollee's State Medicaid program paid the Medicare premiums, and the Census division in which the enrollee resided.
We examined the rates of complications within ninety days after primary and revision total knee replacement for all patients; the complications included death from all causes, readmission to an acute care hospital for additional knee surgery, manipulation under anesthesia, pulmonary embolus (both fatal and nonfatal), pneumonia requiring hospitalization, acute myocardial infarction, and deep wound infection (defined as that occurring after discharge from the index admission and requiring readmission to the hospital). The performance of manipulation with the patient under anesthesia was ascertained by searching for a surgeon's claim with a CPT code of 27570, and the occurrence of pulmonary embolus was ascertained by searching for a hospital claim with an ICD-9-CM code of 415.1-415.19. Acute myocardial infarction and pneumonia were defined with use of ICD-9-CM codes from hospital claims on the basis of algorithms developed by the Agency for Healthcare Research and Quality10. The overall rates of adverse events in the ninety days after total knee replacement were estimated with use of life-table techniques. Death censored all other outcomes.
We used proportional hazards models to examine the relationship between demographic factors and the occurrence of these complications in the ninety days after primary and revision total knee replacement. The factors examined included age-group, gender, race, and Medicaid supplementation status as described above. Again, only enrollees coded as black or white were retained for the proportional hazards models. In addition, an adaptation of the Charlson index of comorbidity was computed with use of inpatient claims data for the index admission11,12. The Charlson index scores were dichotomized at zero versus one or more. For primary procedures, a diagnosis of rheumatoid arthritis also was examined as a potential risk factor. For revision procedures, age-groups of eighty years and greater were combined because the sample was small. The analyses were adjusted for the total number of total knee replacements performed annually in Medicare beneficiaries at the hospital and by the surgeon.
Standardized mortality ratios were used to compare the observed numbers of deaths within ninety days after primary and revision total knee replacements with the expected number of deaths, which were determined with use of mortality rates computed from the 5% United States Medicare sample. From this population, the 1.8 million enrollees who were alive on January 1, 2000 (the first day of our study period) were extracted. Every member of this standard population was then randomly allocated the surgery date of one of the patients who had undergone primary or revision total knee replacement, and that date was used as the starting date for calculating death rates. Members of the standard population who had died before the allotted starting date, or who were not sixty-five years old or more on that date, were dropped. Similarly, members who were enrolled in a health maintenance organization, who were not residents of the United States, or who were not enrolled in both parts of Medicare were dropped because we had excluded these individuals from our total knee replacement cohorts. This process left 1.3 million members of the standard population, who were then divided into seventy-two strata according to age-group, gender, race, and Medicaid buy-in status. Stratum-specific ninety-day death rates were calculated, and the rates were applied to the primary and revision total knee replacement cohorts. Approximate confidence intervals for the standardized mortality ratios were calculated with use of the formulas of Rothman and Boice13.
With use of our case definition criteria, 124,986 Medicare beneficiaries underwent unilateral primary total knee replacement and 11,726 underwent revision total knee replacement between January 1 and December 31, 2000 (Table I). Approximately one-third of each cohort were men, and approximately one-half were seventy-five years old or more. Eight percent (10,388) of the 124,986 beneficiaries who underwent primary total knee replacement and 10% (1117) of the 11,726 who underwent revision total knee replacement qualified for Medicaid supplementation, indicating poverty-level income.
Rates of Total Knee Replacement
In general, the utilization rates for both primary and revision total knee replacement were higher for women than for men, although the female-to-male rate ratios declined with age (Table II). For both primary and revision total knee replacement, the highest utilization rates were for the seventy-five to seventy-nine-year-old age-group among both men and women. There was nearly a twofold difference in the rates of both primary and revision total knee replacement between the lowest and highest rates over Census divisions (see Appendix). The rank ordering of Census divisions by rates for total knee replacement was nearly identical for both primary and revision total knee replacement. The West North Central division had the highest annual rates of both primary total knee replacement (68.2 per 10,000) and revision total knee replacement (6.3 per 10,000).
Poisson regression was used to evaluate the independent associations between demographic characteristics (age, gender, race, Medicaid supplementation, and Census division) and the incidence rates of unilateral primary and revision total knee replacement (Table III). The analysis yielded rate ratios that were adjusted for the other demographic factors in the table. The rate ratios for primary and revision total knee replacement were highest for the seventy-five to seventy-nine-year-old age-group. Women had higher rate ratios for both procedures, particularly primary total knee replacement (rate ratio, 1.41; 95% confidence interval, 1.37 to 1.45). Compared with whites, blacks were less likely to undergo either primary total knee replacement (rate ratio, 0.73; 95% confidence interval, 0.69 to 0.78) or revision total knee replacement (rate ratio, 0.89; 95% confidence interval, 0.81 to 0.99). Compared with individuals who were not receiving Medicaid supplementation, those who were receiving Medicaid supplementation were less likely to undergo either procedure, particularly primary total knee replacement (rate ratio, 0.74; 95% confidence interval, 0.70 to 0.78).
In the first ninety days following unilateral primary total knee replacement, 0.7% of patients died, 1.4% had pneumonia requiring hospitalization, and <1% had a pulmonary embolus, a deep wound infection, an acute myocardial infarction, or required readmission to the hospital for additional knee surgery. The incidence of these outcomes was generally somewhat higher following revision total knee replacement (Table IV).
In proportional hazards models, increasing age was associated with higher rates of death, pulmonary embolus, pneumonia, and myocardial infarction as well as with a strikingly lower rate of manipulation following primary total knee replacement (see Appendix). Women had lower rates than men did for all of the studied outcomes except for pulmonary embolus; the reductions were particularly pronounced for death (rate ratio, 0.5; 95% confidence interval, 0.5 to 0.6), deep wound infection (rate ratio, 0.7; 95% confidence interval, 0.6 to 0.8), pneumonia (rate ratio, 0.7; 95% confidence interval, 0.6 to 0.7), and myocardial infarction (rate ratio, 0.6; 95% confidence interval, 0.5 to 0.6). Compared with whites, blacks had a tendency toward higher mortality (rate ratio, 1.4; 95% confidence interval, 1.0 to 1.8) as well as higher rates of deep wound infection (rate ratio, 1.5; 95% confidence interval, 1.0 to 2.1) and manipulation (rate ratio, 1.4; 95% confidence interval, 1.2 to 1.7). Individuals who had one or more medical comorbidities or who received Medicaid supplementation had higher rates of all of the outcomes except manipulation.
Similar analyses of postoperative complications following revision total knee replacement are also detailed in the Appendix. Greater age was associated with higher rates of mortality and pneumonia requiring hospitalization and lower rates of manipulation. Patterns associated with gender, race, and comorbidity were broadly similar to those for primary total knee replacement. In contrast to the findings after primary total knee replacement, individuals receiving Medicaid supplementation did not have increased rates of additional knee surgery following revision total knee replacement.
The ninety-day standardized mortality ratio after primary total knee replacement was strikingly low (0.6; 95% confidence interval, 0.6 to 0.7), indicating 40% fewer deaths than would be expected in the unselected Medicare population with the same demographic structure. After revision total knee replacement, the ninety-day standardized mortality ratio was 0.9 (95% confidence interval, 0.7 to 1.0), suggesting that the number of deaths after revision was about as expected.
Using United States Medicare data for the year 2000, we documented that the rates of utilization of primary and revision total knee replacement vary by age, gender, race, and geographic region. The rate of primary total knee replacement was about ten times higher than the rate of revision total knee replacement. The rates of both procedures increased with age after sixty-five years and peaked in the seventy-five to seventy-nine-year-old age-group. The rates of both procedures were also higher in women than in men, in whites than in blacks, and in those with higher incomes than in those who received Medicaid supplementation. There was nearly a twofold rate gradient across Census divisions, with the highest utilizations in the West North Central and Mountain regions.
Overall, the risks of adverse outcomes in the ninety days after both primary and revision total knee replacement are very low. Indeed, the risk of mortality after unilateral primary total knee replacement was substantially lower than the risk facing unselected Medicare patients with the same demographic structure. Increasing age was most strongly associated with higher rates of death and pneumonia. The presence of medical comorbidity and Medicaid supplementation were most consistently associated with higher rates of adverse outcomes.
Previous population-based research on the complications of total knee replacement is limited. The ninety-day mortality rate following primary total knee replacement in our study was 30% lower than that in a similar cohort managed with primary total hip replacement in the Medicare population; after revision total knee replacement, the ninety-day mortality rate was less than half of that after revision total hip replacement14. In contrast, the risk of deep wound infection following either primary or revision total knee replacement was nearly twice as high as that after the corresponding total hip replacement procedures14. Other studies of total knee replacement were restricted to the inpatient period and could not identify complications that occurred following discharge from the hospital5.
Lower rates of utilization of medical services by blacks than by whites have been documented for total knee replacement1,3,8,15-17 as well as for cardiac, renal, and other procedures18-21. In previous studies on the rates of total hip replacement, we noted that blacks had lower rates of surgery compared with whites14,22,23. The findings of the current study show similar trends for total knee replacement. This difference in utilization rates is unlikely to be due to differences in the prevalence of osteoarthritis between whites and blacks3,24. The differences across Census regions documented in the present study have been noted in previous studies on total hip replacement and total knee replacement14,25. These differences remain inadequately explained. They could arise from regional differences in the underlying epidemiology of osteoarthrosis, patient preferences, surgical capacity, physician preferences, or access to services. It is difficult to determine the appropriate rate, and it is simplistic to assume that the highest rates are excessive. Data from Ontario suggest that the need for such procedures may be higher in high-rate areas4.
Two additional observations should be noted regarding the utilization of total knee replacement. First, socioeconomic status has a substantial effect on the utilization of total knee replacement after adjustment for race, gender, and age. Individuals whose income level was low enough to qualify for Medicaid supplementation were much less likely to undergo total knee replacement than individuals who did not receive Medicaid supplementation were. This income disparity also has been noted with regard to the utilization of total hip replacement14,23. Second, women had higher rates of utilization of total knee replacement than men did, particularly with respect to primary unilateral procedures. This phenomenon can be explained by the higher prevalence of osteoarthritis of the knee among women who are sixty-five years of age and older24,26. The female-to-male rate ratios decreased with age, so that women who were more than ninety years old were less likely to have a total knee replacement than their male counterparts were (Table II). Older women may be more likely than older men to be living alone or to otherwise lack a supportive social network, and this might explain an unwillingness to undergo total knee replacement.
The highest complication rates following total knee replacement were associated with increasing age, male gender, black race, presence of medical comorbidity, and Medicaid supplementation. These factors are similar to those that have been associated with adverse outcomes following total hip replacement14. Additional research should investigate whether the worse outcomes in blacks and the poor are due to patient-related factors or to the quality of care at the hospitals where total knee replacement is performed.
An important limitation of the present study was our inability to identify individuals in the Medicare population who had preexisting knee osteoarthritis and thus were at risk for primary total knee replacement (or who had had previous primary total knee replacement and thus were at risk for revision total knee replacement). Also, Medicare claims data provide sparse clinical detail, making it difficult, for example, to distinguish a complex revision from a simple revision. The present study had several strengths, including a population-based sample (eliminating referral bias), a large-enough sample size to estimate the rate of infrequent outcomes, and complete follow-up.
Tables presenting total knee replacement rates by Census division and ninety-day outcomes following primary and revision total knee replacement are available with the electronic versions of this article, on our web site at jbjs.org (go to the article citation and click on “Supplementary Material”) and on our quarterly CD-ROM (call our subscription department, at 781-449-9780, to order the CD-ROM). ▪
NOTE: The authors thank Robert A. Lew, PhD, for contributing to the design of the study.
Investigation performed at the Section of Clinical Sciences and the Division of Rheumatology, Immunology and Allergy, the Robert Brigham Arthritis and Musculoskeletal Clinical Research Center, Boston, Massachusetts; the Department of Orthopaedic Surgery, Brigham and Women's Hospital, Harvard Medical School, Boston, Massachusetts; the Department of Biostatistics, Boston University School of Public Health, Boston, Massachusetts; the Departments of Medicine and Community Medicine, Dartmouth Medical School, Lebanon, New Hampshire; and the Musculoskeletal Health and Arthritis Program, Toronto Western Hospital, University Health Network, University of Toronto, Toronto, Canada
In support of their research or preparation of this manuscript, one or more of the authors received grants or outside funding from the National Institutes of Health (P60 AR 47782 and K24 AR 02123). None of the authors received payments or other benefits or a commitment or agreement to provide such benefits from a commercial entity. No commercial entity paid or directed, or agreed to pay or direct, any benefits to any research fund, foundation, educational institution, or other charitable or nonprofit organization with which the authors are affiliated or associated.
1. , Freund DA, Heck DA, Dittus RS, Paul JE, Wright J, Coyte P, Holleman E, Hawker G. Demographic variation in the rate of knee replacement: a multi-year analysis. Health Serv Res. 1996;31: 125-40.
2. Agency for Healthcare Research and Quality. HCUPnet, healthcare cost and utilization project. Rockville, MD: Agency for Healthcare Research and Quality; 2002. http://www.ahrq.gov.
3. , Song J, Manheim LM, Chang RW. Racial disparities in joint replacement use among older adults. Med Care. 2003;41: 288-98.
4. , Wright JG, Coyte PC, Williams JI, Harvey B, Glazier R, Wilkins A, Badley EM. Determining the need for hip and knee arthroplasty: the role of clinical severity and patients' preferences. Med Care. 2001;39: 206-16.
5. , Purves HR, Guller U, Toth AP, Vail TP, Pietrobon R. Provider volume of total knee arthroplasties and patient outcomes in the HCUP-nationwide inpatient sample. J Bone Joint Surg Am. 2003;85: 1775-83.
6. , Barrett J, Mahomed NN, Baron JA, Wright RJ, Losina E. Association between hospital and surgeon procedure volume and the outcomes of total knee replacement. J Bone Joint Surg Am. 2004;86: 1909-16.
7. , Garfinkel SA, McQuay LJ, Heck DA, Wright JG, Dittus R, Lubitz RM. The effect of hospital volume on the in-hospital complication rate in knee replacement patients. Health Serv Res. 1998;33: 1191-210.
8. , Weinstein JN, Sporer SM, Wennberg JE. Racial, ethnic, and geographic disparities in rates of knee arthroplasty among Medicare patients. N Engl J Med. 2003;349: 1350-9.
9. , Drake BG, Heck DA, Dittus RS. Patient outcomes following tricompartmental total knee replacement. A meta-analysis. JAMA. 1994;271: 1349-57.
11. , 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.
12. , Roos LL, Jollis JG. Adapting a clinical comorbidity index for use with ICD-9-CM administrative data: differing perspectives. J Clin Epidemiol. 1993;46: 1075-90.
13. , Day NE. Statistical methods in cancer research. Volume 2, The design and analysis of cohort studies. Lyon: International Agency for Research on Cancer; 1987.
14. , 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.
15. , Oishi CS, Gialamas GG. Extreme variations in racial rates of total hip arthroplasty for primary coxarthrosis: a population-based study in San Francisco. Ann Rheum Dis. 1995;54: 107-10.
16. , Gornick M. Differences by race in the rates of procedures performed in hospitals for Medicare beneficiaries. Health Care Financ Rev. 1994;15: 77-90.
17. , Center for the Evaluative Clinical Sciences. The Dartmouth atlas of musculoskeletal health care. American Hospital Association. Chicago: Press; 2000.
18. , Ayanian JZ. Racial disparities in medical care. N Engl J Med. 2001;344: 1471-3.
19. , Udvarhelyi IS, Gatsonis CA, Pashos CL, Epstein AM. Racial differences in the use of revascularization procedures after coronary angiography. JAMA. 1993;269: 2642-6.
20. , Franks P, Gold MR, Clancy CM. Inequality in quality: addressing socioeconomic, racial, and ethnic disparities in health care. JAMA. 2000;283: 2579-84.
21. , Franks P, Gold MR, Clancy CM. Inequalities in racial access to health care. JAMA. 2000;284: 2053.
22. , Barrett J, Katz JN, Liang MH. Total hip arthroplasty: use and select complications in the US Medicare population. Am J Public Health. 1996;86: 70-2.
23. , Barrett J, del Rincon I, Cornell JE, Phillips CB, Katz JN. Disparity in total hip replacement affecting Hispanic Medicare beneficiaries. Med Care. 2002;40: 451-60.
24. , Linder GF, Renner JB, Fryer JG. The impact of arthritis in rural populations. Arthritis Care Res. 1995;8: 242-50.
25. , Hollenberg JP, Szatrowski TP, Johanson NA, Mancuso CA, Charlson ME. Geographic variations in the rates of elective total hip and knee arthroplasties among Medicare beneficiaries in the United States. J Bone Joint Surg Am. 1992;74: 1530-9.
26. , Naimark A, Anderson J, Kazis L, Castelli W, Meenan RF. The prevalence of knee osteoarthritis in the elderly. The Framingham Osteoarthritis Study. Arthritis Rheum. 1987;30: 914-8.