Total hip arthroplasty (THA) is a remarkably successful and widely used procedure,1,2 yet despite an equal prevalence of hip arthritis in blacks and whites,3 blacks undergo THA markedly less often than whites, a disparity in THA utilization that has persisted for decades.4 The explanation for this disparity is not known. Blacks are more likely to undergo arthroplasty in hospitals with a low volume of arthroplasties, in which functional outcomes and patient satisfaction are worse.5,6 Blacks are more likely to prefer conservative and nontraditional arthritis treatments and to have lower expectations of arthroplasty outcomes than whites.7-10 However, whether these expectations and preferences are appropriate is not known because few studies of arthroplasty outcomes include race in their analysis, and those that have analyzed race demonstrate that differences between blacks and whites are small.11-14 Complicating matters further, the effect of socioeconomic factors such as poverty and education is not always included in the analysis of THA outcomes.12-14
In a previous study, we demonstrated that Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) pain and function after total knee arthroplasty (TKA) are similar for blacks and whites from neighborhoods with little poverty, whereas notable racial disparities exist among patients from impoverished neighborhoods.15 These findings demonstrate the importance of including socioeconomic factors at both the individual and community level in the analysis of outcomes by race. However, outcomes differ for THA versus TKA, because up to 30% of patients undergoing TKA may not be satisfied with the results of their surgery or have residual pain, whereas over 90% of patients undergoing THA report satisfaction with their results.1,16,17
Socioeconomic measures at the community level can be determined through the analysis of census tracts. These small population aggregates are designed to be homogenous with regard to living conditions and economic status, factors that have been shown to contribute to health outcomes.18-20 Individual patient addresses can be linked to corresponding census tracts by geocoding, permitting analysis of socioeconomic variables at the community level. A multilevel modeling approach can then be used to quantify and map socioeconomic disparities at both the individual and community level across census tracts, revealing measurable gradients in health outcomes.20 Previous studies demonstrate poorer health outcomes in communities with >20% of the community below the poverty line and that blacks living in these communities may be disproportionately affected.20,21
The purpose of this study was to determine whether neighborhood socioeconomic factors have a differential effect in blacks and whites on WOMAC pain and function 2 years after undergoing THA at the same high-volume hospital.
We used prospectively acquired institutional THA registry data in this retrospective study, which was approved by our institutional review board. All patients undergoing THA in a single, high-volume orthopaedic hospital between May 1, 2007, and February 25, 2011, were asked to participate in the registry, with registry data collected at baseline and 2 years after THA. More than 82% of the patients who were asked to participate consented to be in the registry and provided data. We included adults aged above 18 years who provided baseline and 2-year data, had identified race, and had a geocodable address in New York, New Jersey, or Connecticut. We selected this geographic area to exclude patients traveling from remote locations because they were likely to be wealthy socioeconomic outliers. Patients were excluded if they had International Classification of Diseases (ICD)-9 codes for fracture, were undergoing revision or bilateral THA, or had contralateral THA or TKA within 2 years of the index surgery so that the 2-year outcome data would accurately reflect the index surgery. We excluded Asians, Pacific Islanders, and Native Americans because of their small numbers within our study cohort.
Baseline data collected on all registry patients included age, sex, body mass index (BMI), ethnicity (ie, non-Hispanic or Hispanic), race, insurance status (ie, Medicare, Medicaid, or other), and education (ie, some college or above or no college). Administrative data included Charlson-Deyo comorbidity scores calculated from ICD-9 codes,22 patient address, and race (if available) when not self-identified by the patient. Patient-reported outcome measures included the hospital THA Expectations Survey23 and the Hip Osteoarthritis Outcomes Survey (HOOS), from which we derived the WOMAC pain and function scores.24-27 Our hospital's THA Expectations Survey is a validated instrument that assesses patient expectations before surgery that are specific to hip replacement, including relief of pain, walking, and ability to perform essential activities. The survey is scored on a scale of 1 to 100, higher is better, and the minimum clinically important difference (MCID) is considered to be 7.23,28 The HOOS and WOMAC are validated, lower extremity-specific surveys that can be used to assess pain and function after THA. WOMAC pain and function scores, derived from the HOOS, are scored on a scale of 1 to 100, higher is better, with an MCID of 10.24,29
We used the Geographic Information Systems (ArcGIS) mapping program to link individual patient addresses to specific census tracts. Census tract-level variables were derived from the American Community Survey/US Census.20,30 We screened multiple census tract variables for use in our model as an area-based socioeconomic measure, including the percent of the population living below the poverty level (“census tract poverty”), the percent with Medicaid insurance coverage (“census tract Medicaid coverage”), percent living alone, percent single mothers, percent older than 65, percent black or Hispanic, median household income, and the Gini coefficient, a measure of income inequality based on income distribution.
We used descriptive statistics to summarize baseline and 2-year patient characteristics, census tract level variables, and differences between white and black patients. Baseline characteristics were compared for patients by availability of geocodable addresses, self-reported race, and 2-year WOMAC pain and function scores using a t-test or Wilcoxon rank-sum test for continuous variables and a chi-square or Fisher exact test for categoric variables. No imputation of missing data was done, because less than 3% was missing (if missing at random is assumed). Census tract level variables were summarized for all census tracts using descriptive statistics. We calculated the correlation between census tract level variables using Spearman correlation coefficients.
A multivariable linear mixed-effect model with a random intercept for each census tract was first performed by including only patient-level variables as predictors of 2-year WOMAC pain and function. In univariate analysis, we explored age, sex, BMI, race, ethnicity, education, Charlson-Deyo comorbidities, insurance status, hospital Expectations score, and baseline WOMAC pain and function for associations with 2-year WOMAC pain and function, and we included any terms significant at the P < 0.1 level in our model to try to account for potentially confounding variables. For multivariable analysis of predictors of pain at 2 years, sex and Charlson-Deyo comorbidities were not statistically significant if included in the model, and compared with other candidate models, the model excluding sex and Charlson-Deyo comorbidities had the lowest Akaike information criterion. However, both variables were kept in the final model because they were variables of clinical interest. For multivariable analysis of predictors of function at 2 years, sex was not statistically significant if included in the model and compared with other candidate models, the model excluding sex had the lowest Akaike information criterion. However, again, sex was kept in the final model because it was a variable of clinical interest.
Models for pain and function were then expanded to include census tract level variables, added in one at a time. If a census tract level variable was statistically significant in the model, the interaction between race and that variable was assessed. Among these census tract level variables, percent Medicaid coverage at the census tract level (dichotomized as ≤10% and >10%) was not found to be significant (P = 0.067) when added to the model with 2-year WOMAC pain as the dependent variable and was found to be statistically significant (P = 0.028) when added to the model with 2-year WOMAC function as the dependent variable. The interaction between race and percent Medicaid coverage was assessed for both models. We used percent Medicaid coverage, which was statistically significant in our model, as our census tract variable representing community deprivation based on the strength of the interaction. Percent below the poverty line was not significant in this model.
There were 4,571 THA cases in the cohort from New York, New Jersey, and Connecticut. We excluded 48 revisions, 262 bilateral procedures, 99 patients whose self-identified race was neither black nor white, and 12 who lacked geocodable addresses; some had multiple exclusion criteria. No statistically significant difference was found in any patient characteristics between patients with and without geocodable addresses. Patients with missing race were all Hispanic and, compared with patients not missing race, were younger, with worse pain and function at baseline, and were more likely to have Medicaid insurance. Patients with missing 2-year WOMAC pain scores were older, less college educated, and more often categorized as American Society of Anesthesiology (ASA) class III to IV. Patients with missing 2-year WOMAC function scores were older, less college educated, and had worse WOMAC function at baseline.
Our final cohort included 4,170 patients, of whom 4,025 (97%) were white and 145 (3%) were black (see Table 1, Supplemental Digital Content 1, Characteristics of the Cohort, https://links.lww.com/JAAOS/A127). Compared with whites, blacks were younger, more likely to be female, and had higher BMI, higher hospital expectations scores, and more comorbidities. Blacks had lower educational attainment and were more likely to be insured by Medicaid. Mean WOMAC pain and function scores at baseline and 2 years were seven points lower (worse) for blacks than for whites (P < 0.0001). Compared with blacks and whites from wealthier neighborhoods, both blacks and whites from impoverished neighborhoods with >10% Medicaid coverage had worse baseline pain (blacks 52.8 versus 43.8, P = 0.02; whites 55 versus 54.8, P = 0.004) and worse baseline function (blacks 49.06 versus 40.05, P = 0.003; whites 58.8 versus 50.8, P < 0.001).
Blacks and whites achieved similar improvements in pain and function from baseline to 2 years. The change in WOMAC pain was 42 ± 24 for blacks versus 40 ± 19 for whites (P = 0.14), and the change in WOMAC function was 42 ± 24 for blacks versus 41 ± 19 for whites (P = 0.35). More blacks than whites lived in neighborhoods with 20% or greater of the population below the poverty level (30% versus 3%; P < 0.0001), and more blacks than whites lived in neighborhoods with 10% or greater of the population with Medicaid insurance coverage (71% versus 23%; P < 0.0001).
Census Tract-level Variables
Patients in the registry came from a total of 1,916 census tracts in New York, New Jersey, and Connecticut (of a total of 7,762 census tracts in these three states) (Figure 1). Of these census tracts, 1,009 (53%) were linked to only 1 registry patient address, 753 (39%) were linked to 2 to 4 patient addresses, and the remaining 154 (8%) were linked to 5 to 29 patient addresses (see Appendix 1, Supplemental Digital Content 2, Census Tract Distribution, https://links.lww.com/JAAOS/A128). Ninety-four percent of included census tracts were categorized as urban. The median census tract population was 3% black (range, zero to 100%) and 8% Hispanic (range, zero to 92%). Median census tract family income was $86,959 (range, $13,859 to $250,000) (see Appendix 2, Supplemental Digital Content 3, Census Tract Data Summary, https://links.lww.com/JAAOS/A129). Strong correlations were found between census tract Medicaid coverage and other census tract level variables, including a strong positive correlation with percent living below the poverty level (rho = 0.69; P < 0.001) and a strong negative correlation with median household income (rho = −0.79; P < 0.001) (see Appendix 3, Supplemental Digital Content 4, Correlation Between Census Tract Variables and Percent of the Population With Medicaid Coverage, https://links.lww.com/JAAOS/A130).
Models With Patient-level Data Only
Multivariate analysis of patient-level characteristics demonstrated that black race, Medicaid insurance, lower hospital Expectations score, and worse baseline WOMAC pain were associated with worse WOMAC pain scores 2 years after THA, although the differences were small (see Table 2, Supplemental Digital Content 5, Multivariate Analysis of Predictors of Pain and Function at 2 Years, https://links.lww.com/JAAOS/A131). Black race, Medicaid insurance, greater age, higher BMI, more comorbidities, and worse baseline WOMAC function were associated with worse WOMAC function scores 2 years after THA (see Table 2, https://links.lww.com/JAAOS/A131).
Census Tract Level Data
When we incorporated the percent of the population with Medicaid coverage at the census tract level (≤10% versus >10%) into our models, we found that >10% census tract Medicaid coverage was associated with worse 2-year WOMAC function scores (P = 0.028), with a trend toward worse WOMAC pain scores as well (P = 0.067). Black race remained associated with worse 2-year WOMAC pain (P = 0.02) and function (P = 0.008) after incorporating census tract Medicaid coverage into our model (see Table 3, Supplemental Digital Content 6, Effect of Adding Census Tract Variables to Individual-level Data, https://links.lww.com/JAAOS/A132).
Interaction of Race and Census Tract Data
Only a mild interaction was found between race and census tract Medicaid coverage when modeling their combined effect on 2-year WOMAC pain, but a very strong interaction was found between the two variables in estimates of 2-year WOMAC function (Figures 2 and 3). In communities with progressively higher census tract Medicaid coverage (see Table 4, Supplemental Digital Content 7, WOMAC Pain and Function 2 Years After THA, https://links.lww.com/JAAOS/A133), we observed a stepwise worsening of 2-year WOMAC function in all patients, which was much more pronounced among blacks than whites. For example, with the other variables in the model held constant at their mean values (WOMAC function at baseline = 49; age at surgery = 65; BMI = 28/kg/m2; hospital Expectations score = 83; sex = female; comorbidities = zero; insurance = Medicaid; education = college and above), estimated 2-year WOMAC function scores were only 1.99 ± 1.44 lower among blacks compared with whites living in communities with zero to 10% census tract Medicaid coverage (P = 0.17), but 6.45 ± 2.24 lower in communities with >30% to 40% census tract Medicaid coverage (P = 0.004). Looked at in another way, comparing blacks in census tracts with Medicaid coverage >30% to 40% with those in communities with zero to 10% census tract Medicaid coverage, WOMAC function scores were predicted to be five points lower for blacks living in more deprived communities (>30% to 40% Medicaid coverage, whites = 85.96 versus blacks = 80.42), whereas WOMAC function scores were only 1 point lower among blacks living in less deprived communities (zero to 10% Medicaid coverage, whites = 87.95 versus blacks = 86.87) (Figure 3).
In this study, we have demonstrated that blacks and whites from wealthier neighborhoods with little Medicaid coverage have similar pain and function 2 years after THA. However, using high levels of community Medicaid coverage as an indicator of community economic deprivation, blacks from neighborhoods with high levels of Medicaid coverage have worse functional outcomes at 2 years after THA compared with whites from similar neighborhoods, despite undergoing surgery at the same high-volume hospital. THA is generally a successful operation, and poor outcomes are much rarer than for TKA.1,31 Nonetheless, we demonstrate that socioeconomic factors magnify racial disparities even for the outcome of this reliable procedure.1,17 Although the difference between the best and worst outcomes in our study (in absolute terms) was small, and less than the MCID, this is consistent with the findings of other analyses,12-14 the effect of community Medicaid coverage was “dose-dependent,” which helps to validate our results. Our findings are consistent with other studies that demonstrate an association between community deprivation and adverse outcomes after TKA and THA, along with other health conditions such as diabetes care and all-cause mortality.15,18,21,32,33 In our study, we also demonstrate that although the degree of improvement in pain and function is similar between blacks and whites, being black, having Medicaid insurance coverage (at the individual level), and having worse pain and function at baseline all independently predict worse pain and function at 2 years after THA.
Blacks in our study had worse pain and function at baseline than white patients despite presenting at a younger age and had more comorbidities than whites at the time of surgery.14,15 Interestingly, whites from impoverished neighborhoods also had worse pain and function at baseline than whites from wealthier neighborhoods, which suggests that poverty, not race, may mediate delays in seeking surgical care. Black patients in our study actually experienced greater improvements in WOMAC pain and function after surgery (though not statistically different) than whites did; however, because their baseline pain and function was lower, they never achieved comparable postoperative results. This finding suggests that future efforts to improve racial disparities in total hip replacement outcomes should focus on patient education regarding the benefits of total hip replacement, and recruitment of poor black patients with osteoarthritis at an earlier stage in their disease.
Access to health insurance can mitigate some of the racial disparities in arthroplasty utilization, as was demonstrated after healthcare reform in Massachusetts.34 However, even in countries with universal insurance access, both black and impoverished patients seek arthroplasty later than those who are neither.35 The fact that more blacks than whites live in poverty further magnifies the effect of neighborhood poverty on racial disparities in health outcomes.36 We selected census tract Medicaid coverage as the area-based socioeconomic measure in this study because, of those area-based poverty measures assessed, it performed the best in our model. To our surprise, census tract poverty was not associated with poor THA outcomes in our model. This suggests that factors related to the relationship between community poverty and the healthcare system (ie, insurance coverage), rather than poverty alone, may be more important in mediating racial disparities in THA outcomes. Greene et al37 demonstrate that physicians are markedly less likely to participate in Medicaid in areas where the poor are nonwhite and in areas that are racially segregated.
Given the similar prevalence of hip osteoarthritis in blacks and whites, and despite blacks and whites deriving the same degree of improvement after THA, do poorer outcomes for blacks contribute to the continued underuse of this effective procedure?3,4 Given that the disparity in THA utilization persists even among blacks and whites with equally severe arthritis, this question could be answered in part by blacks' greater preference for conservative therapies, including physical therapy and alternative therapies such as prayer and massage.9 Others have reported that blacks have different expectations for arthroplasty that include expectations of slower recuperation and greater postoperative pain.10 However, in our study, despite poorer scores for pain and function at baseline, blacks had higher expectations for surgery compared with whites. It is possible that both of these factors contribute delays in seeking surgical care, contributing to poorer outcomes.
Our study has some limitations. Patients with poor outcomes are less likely to return follow-up questionnaires.38 In our study, patients without 2-year follow-up were older, in poorer health, less educated, and had poorer baseline WOMAC function. This may have skewed the population toward patients with better outcomes and blunted the magnitude of our findings. Similarly, because all THAs were performed in a high-volume hospital, the results may not be generalizable; however, this is most likely to have decreased the magnitude of our findings.6 Our cohort, despite being the largest published US cohort comparing patient-reported outcomes after THA in blacks and whites, is nonetheless limited by the small number of blacks in our cohort, the small number of patients residing in impoverished communities, and the small number of cases with poor outcomes. Further limiting generalizability, the patients in our cohort were highly educated, with 68.69% of whites and 58.74% of blacks attaining education at or above the college level, and few resided in rural communities. Nonetheless, we were able to demonstrate a “dose dependent” interaction between race and community level deprivation. Although we were concerned that Medicaid coverage at the individual level might be a proxy for physician trainees (surgical residents and fellows) performing the surgery, this was not the case for TKA performed at the same institution, where Medicaid insurance was not a risk factor for poorer outcomes.15 Of note, TKA is considered a more technically complex procedure than THA. However, surgeon volume is a predictor of THA dislocation and revision,39 and we cannot rule out the possibility that surgeon experience played a role. Medicaid insurance was also used as a surrogate for socioeconomic status, and our modeling revealed differences in community Medicaid coverage and community poverty. Although our previous study of differences in TKA outcomes between blacks and whites revealed a strong association between community poverty and arthroplasty outcomes, this study does not. This finding suggests that additional factors related to census tract Medicaid coverage are not captured by other area-based measures of poverty, as others have also reported.40 Our use of ICD-9 codes to determine comorbidities could have led to inaccuracies and, as in all observational studies, unmeasured confounders may have biased our results.
In summary, we demonstrate a notable relationship between socioeconomic status and THA outcomes, despite the notable benefit after THA seen for both blacks and whites. Blacks living in neighborhoods with little Medicaid coverage have similar functional outcomes after THA compared with whites, but as the community level Medicaid coverage increases, functional outcomes for blacks worsen to a much greater extent than for whites, even when all patients have access to care at the same high-volume hospital.
References printed in bold type are those published within the past 5 years.
1. Learmonth ID, Young C, Rorabeck C: The operation of the century: Total hip replacement. Lancet 2007;370:1508-1519.
2. Kurtz S, Ong K, Lau E, Mowat F, Halpern M: Projections of primary and revision hip and knee arthroplasty in the united states from 2005 to 2030. J Bone Joint Surg Am 2007;89:780-785.
3. Jordan JM, Helmick CG, Renner JB, et al.: Prevalence of hip symptoms and radiographic and symptomatic hip osteoarthritis in African Americans and Caucasians: The Johnston county osteoarthritis project. J Rheumatol 2009;36:809-815.
4. Singh JA, Lu X, Rosenthal GE, Ibrahim S, Cram P: Racial disparities in knee and hip total joint arthroplasty: An 18-year analysis of national medicare data. Ann Rheum Dis 2014;73:2107-2115.
5. Katz JN, Losina E, Barrett J, et al.: 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:1622-1629.
6. Katz JN, Phillips CB, Baron JA, et al.: Association of hospital and surgeon volume of total hip replacement with functional status and satisfaction three years following surgery. Arthritis Rheum 2003;48:560-568.
7. Groeneveld PW, Kwoh CK, Mor MK, et al.: Racial differences in expectations of joint replacement surgery outcomes. Arthritis Rheum 2008;59:730-737.
8. Ibrahim SA, Siminoff LA, Burant CJ, Kwoh CK: Differences in expectations of outcome mediate african american/white patient differences in “willingness” to consider joint replacement. Arthritis Rheum 2002;46:2429-2435.
9. Ang DC, Ibrahim SA, Burant CJ, Siminoff LA, Kwoh CK: Ethnic differences in the perception of prayer and consideration of joint arthroplasty. Med Care 2002;40:471-476.
10. Allen KD, Golightly YM, Callahan LF, et al.: Racial and gender differences in willingness to undergo total joint replacement: The Johnston county osteoarthritis project. Arthritis Care Res (Hoboken) 2014;66:1193-1202.
11. Goodman SM, Parks ML, McHugh K, et al.: Disparities in outcomes for African Americans and Whites undergoing total knee arthroplasty: A systematic literature review. J Rheumatol 2016;43:765-770.
12. Allen Butler R, Rosenzweig S, Myers L, Barrack RL: The Frank Stinchfield Award: The impact of socioeconomic factors on outcome after THA: A prospective, randomized study. Clin Orthop Relat Res 2011;469:339-347.
13. Lavernia CJ, Alcerro JC, Contreras JS, Rossi MD: Ethnic and racial factors influencing well-being, perceived pain, and physical function after primary total joint arthroplasty. Clin Orthop Relat Res 2011;469:1838-1845.
14. Lavernia CJ, Villa JM: Does race affect outcomes in total joint arthroplasty? Clin Orthop Relat Res 2015;473:3535-3541.
15. Goodman SM, Mandl LA, Parks ML, et al.: Disparities in TKA outcomes: Census tract data show interactions between race and poverty. Clin Orthop Relat Res 2016;474:1986-1995.
16. Jones CA, Voaklander DC, Johnston DW, Suarez-Almazor ME: Health related quality of life outcomes after total hip and knee arthroplasties in a community based population. J Rheumatol 2000;27:1745-1752.
17. Harris WH, Sledge CB: Total hip and total knee replacement (1). N Engl J Med 1990;323:725-731.
18. Subramanian SV, Chen JT, Rehkopf DH, Waterman PD, Krieger N: Racial disparities in context: A multilevel analysis of neighborhood variations in poverty and excess mortality among black populations in Massachusetts. Am J Public Health 2005;95:260-265.
19. Subramanian SV, Chen JT, Rehkopf DH, Waterman PD, Krieger N: Comparing individual- and area-based socioeconomic measures for the surveillance of health disparities: A multilevel analysis of Massachusetts births, 1989-1991. Am J Epidemiol 2006;164:823-834.
20. Krieger N, Chen JT, Waterman PD, Rehkopf DH, Subramanian SV: Painting a truer picture of US socioeconomic and racial/ethnic health inequalities: The public health disparities geocoding project. Am J Public Health 2005;95:312-323.
21. Chen JT, Rehkopf DH, Waterman PD, et al.: Mapping and measuring social disparities in premature mortality: The impact of census tract poverty within and across boston neighborhoods, 1999-2001. J Urban Health 2006;83:1063-1084.
22. Deyo RA, Cherkin DC, Ciol MA: Adapting a clinical comorbidity index for use with ICD-9-CM administrative databases. J Clin Epidemiol 1992;45:613-619.
23. Mancuso CA, Salvati EA, Johanson NA, Peterson MG, Charlson ME: Patients' expectations and satisfaction with total hip arthroplasty. J Arthroplasty 1997;12:387-396.
24. Bellamy N: Instruments to assess osteoarthritis: Current status and future needs. Ann Rheum Dis 1995;54:692-693.
25. Bellamy N: Outcome measurement in osteoarthritis clinical trials. J Rheumatol Suppl 1995;43:49-51.
26. Nilsdotter A, Bremander A: Measures of hip function and symptoms: Harris hip score (HHS), hip disability and osteoarthritis outcome score (HOOS), oxford hip score (OHS), lequesne index of severity for osteoarthritis of the hip (LISOH), and american academy of orthopedic surgeons (AAOS) hip and knee questionnaire. Arthritis Care Res (Hoboken) 2011;63(suppl 11):S200-S207.
27. Nilsdotter AK, Lohmander LS, Klassbo M, Roos EM: Hip disability and osteoarthritis outcome score (HOOS): Validity and responsiveness in total hip replacement. BMC Musculoskelet Disord 2003;4:10.
28. Mancuso CA, Jout J, Salvati EA, Sculco TP: Fulfillment of patients' expectations for total hip arthroplasty. J Bone Joint Surg Am 2009;91:2073-2078.
29. Ehrich EW, Davies GM, Watson DJ, Bolognese JA, Seidenberg BC, Bellamy N: Minimal perceptible clinical improvement with the western ontario and McMaster universities osteoarthritis index questionnaire and global assessments in patients with osteoarthritis. J Rheumatol 2000;27:2635-2641.
30. Krieger N, Chen JT, Waterman PD, Rehkopf DH, Subramanian SV: Race/ethnicity, gender, and monitoring socioeconomic gradients in health: A comparison of area-based socioeconomic measures—the public health disparities geocoding project. Am J Public Health 2003;93:1655-1671.
31. Carr AJ, Robertsson O, Graves S, et al.: Knee replacement. Lancet 2012;379:1331-1340.
32. Hashiguchi Y, Hase K, Ueno H, et al.: Impact of race/ethnicity on prognosis in patients who underwent surgery for colon cancer: Analysis for White, African, and East Asian Americans. Ann Surg Oncol 2012;19:1517-1528.
33. Bass AR, McHugh K, Fields K, Goto R, Parks ML, Goodman SM: Higher total knee arthroplasty revision rates among united states blacks than whites: A systematic literature review and meta-analysis. J Bone Joint Surg Am 2016;98:2103-2108.
34. Hanchate AD, Kapoor A, Katz JN, et al.: Massachusetts health reform and disparities in joint replacement use: Difference in differences study. BMJ 2015;350:h440.
35. Neuburger J, Hutchings A, Allwood D, Black N, van der Meulen JH: Sociodemographic differences in the severity and duration of disease amongst patients undergoing hip or knee replacement surgery. J Public Health (Oxf) 2012;34:421-429.
36. Centers for Disease Control and Prevention (CDC): Racial disparities in total knee replacement among medicare enrollees–united states, 2000-2006. MMWR Morb Mortal Wkly Rep 2009;58:133-138.
37. Greene J, Blustein J, Weitzman BC: Race, segregation, and physicians' participation in medicaid. Milbank Q 2006;84:239-272.
38. Kim J, Lonner JH, Nelson CL, Lotke PA: Response bias: Effect on outcomes evaluation by mail surveys after total knee arthroplasty. J Bone Joint Surg Am 2004;86-A:15-21.
39. Ravi B, Jenkinson R, Austin PC, et al.: Relation between surgeon volume and risk of complications after total hip arthroplasty: Propensity score matched cohort study. BMJ 2014;348:g3284.
40. Foraker RE, Rose KM, Whitsel EA, Suchindran CM, Wood JL, Rosamond WD: Neighborhood socioeconomic status, medicaid coverage and medical management of myocardial infarction: Atherosclerosis risk in communities (ARIC) community surveillance. BMC Public Health 2010;10:632.