We are unable to display your institutional affiliation without JavaScript turned on.
Browse Book and Journal Content on Project MUSE

Download PDF

Racial Disparities in Receiving Total Hip/Knee Replacement Surgery:
The Effect of Hospital Admission Sources

Using a nationally-representative inpatient care dataset (the HCUP National Inpatient Sample from 2002 to 2007) we examined racial disparities in receiving total hip replacement (THR) and total knee replacement (TKR) surgeries. Multivariable logistic regression models revealed that racial minorities were significantly less likely to receive THR or TKR than Whites, controlling for patients' hospital admission source and hospital characteristics. Employing Blinder-Oaxaca decomposition techniques, we found that observed difference in population characteristics explained 55%-67% and 78% of the racial disparities in THR and TKR, respectively. Differences in patients' hospital admission source emerged as the major individual factor associated with these disparities, explaining 57%-77% of racial disparities in THR and 26%-50% of racial disparities in TKR. This study suggests that substantive racial and ethnic disparities exist in utilization of THR and TKR surgery. Observed population characteristics accounted for most of these differences, with hospital admission source being the key factor.

Racial disparity, total joint replacement, hospital admission source, decomposition

Arthritis affects 21% of adults in the United States,1 a prevalence that is projected to rise by 40% in the next 25 years.2 -4 Arthritis is the leading cause of disability and work limitations in the country.5 The estimated economic cost of arthritis in the U.S. was approximately $60 billion in 2001, and is projected to increase to $100 billion by 2020.2 Total joint replacements are important surgical interventions for treating severe arthritis of weight-bearing joints, such as total knee replacement (TKR) and total hip replacement (THR). The literature shows that total joint replacement surgeries can significantly increase quality-adjusted life years, especially for patients that have osteoarthritis associated with significant functional limitations.5 -7 [End Page 135]

Racial disparities in health care utilization have been well-studied in the literature.8 -13 Studies have shown that racial minorities are less likely to receive surgery,5 ,8 -11 even though Whites and minorities are equally clinically appropriate candidates for total joint arthroplasty.12 -13 However, few studies have examined specific factors responsible for racial disparities in receiving major surgeries, and the relative importance of these factors. Skinner et al. showed that geography and sex are major barriers for racial minorities in obtaining surgery.8 Hanchate et al. found that limited insurance coverage and financial constraints explain some of the racial disparities in TKR rates using the longitudinal Health and Retirement Study.14 However, Steel et al. used the same data set and reported that, after taking patients' medical conditions into consideration, only education appeared to be a barrier for receiving surgery, while age, gender, relative poverty, and obesity were unrelated to surgery access.10

Thus, further studies are needed to clarify the specific factors associated with these racial disparities. We used a nationally-representative dataset to estimate racial disparities in obtaining THR/TKR procedures among candidate patients. In contrast to previous studies, we not only examined the effects of patients' socioeconomic characteristics, but also considered the effects of patients' hospital admission source, as well as the characteristics of the admitting hospital, including teaching status, location, ownership status, and bed size, factors that have been shown to be related to access to medical care.15 Particularly, we hypothesized that patient's hospital admission source was an important factor associated with racial disparities in receiving THR and TKR procedures. One reason for this hypothesis is that hospital admission source (such as emergency department vs. regular admission) may indicate severity of illness.11 Possibly, racial minorities have more severe arthritis compared with Whites either due to genetic factors or treatment delays resulting from insufficient access to care. In addition, evidence has shown that racial and ethnic minorities have less health care access than Whites.16 -18 Since THR and TKR procedures are referral-sensitive procedures, minorities' lack of a usual source19 -21 of care may be an important barrier preventing them from receiving timely physician referrals and regular admissions for them. Therefore, differences in admission source may be a key factor associated with racial disparities in receiving THR/TKR procedures.


We used the Nationwide Inpatient Sample (NIS) of the Healthcare Cost and Utilization Project (HCUP) administered by the Agency for Healthcare Research and Quality.22 The NIS is a nationally-representative dataset of US hospital inpatient discharges. The NIS sampled 20% of U.S. community hospitals, which are defined to be "all non-Federal, short-term, general, and other specialty hospitals, excluding hospital units of institutions."22 [p.9] Specialty hospitals such as orthopedic institutions, public hospitals and academic medical centers are included. The NIS contains approximately eight million observations of hospital stays each year, comprising approximately 90% of all hospital discharges in the United States. It also provides detailed information on patient admission source, diagnoses and procedure codes (ICD-9-CM), patients' demographic information (such as age, gender, race, and health insurance status), as well as characteristics of the admitting hospital, such as hospital location and ownership type.

In this study, we employed NIS data from 2002 to 2007 (both inclusive) to examine racial and ethnic disparities (e.g., Whites vs. African Americans and Hispanics, respectively) [End Page 136] in receiving THR (primary ICD9 procedure code = 81.51) or TKR (primary ICD9 procedure code = 81.54) among patients with osteoarthritis or rheumatoid arthritis.23 -25 More specifically, patients with primary diagnosis ICD9 codes of 715.09, 715.15, 715.25, 715.35, or 715.95 were considered as potential candidates for THR, while patients with primary diagnosis ICD9 codes of 714.00, 715.09, 715.16, 715.26, 715.36, or 715.96 were considered as potential patients for TKR (Table 1). Patients with any diagnosis codes for accidental falls were excluded. By using ICD9 codes for specific diagnoses, we selected those patients who were most eligible for the procedures. We included patients aged 25 and older, since few people younger than 25 years old are candidates for these procedures.6 These exclusions left a final sample of 150,525 potentially eligible patients for THR and 366,085 potentially eligible patients for TKR. To examine racial and ethnic disparities in receiving these surgeries, we categorized the racial groups as Whites, African Americans, and Hispanics.*

ICD-9 Diagnose Codes for Patients Who Were Eligible for Total Hip Replacement Procedure (THR) and Total Knee Replacement Procedure (TKR)
Click for larger view
Table 1. 

ICD-9 Diagnose Codes for Patients Who Were Eligible for Total Hip Replacement Procedure (THR) and Total Knee Replacement Procedure (TKR)

[End Page 137]


Outcome variables

Our outcome measures were two binary variables which equaled 1 if patients received the total hip replacement (THR) procedure (i.e., ICD9 procedure code = 81.51) or if patients received total knee replacement (TKR) procedures (i.e. ICD9 procedure code = 81.54), and 0 otherwise. Patients receiving partial replacement surgeries or revisions of surgeries were excluded, since they represented just 8% of all replacement procedures.26

Independent variables

Following previous work, other covariates affecting the likelihood of receiving these procedures were categorized as: 1) factors that influence the patient; 2) factors that influence the physician, 3) health system factors, and 4) contextual factors.11 ,27 -29

Characteristics that can affect the patient's decision to use health care included race, age strata, gender,11 and median household income for patient's ZIP code.28 Factors affecting physicians' decisions to treat included hospital admission sources (emergency department or regular/routine admission sources), the type of admission (whether the procedure was elective or non-elective),11 whether the patient had other chronic diseases, and patients' specific ICD9 primary diagnosis code, which indicated patients' clinical need for the procedures.15 Health system factors included health insurance variables (private health plans, Medicare, Medicaid, self-pay, no charge, and other), and admitting hospital characteristics such as hospital's teaching status, urban/rural location, number of hospital beds (small, medium, and large), and hospital ownership type (general, private, or public owned).* We used the total number of THRs or TKRs at the ZIP code level for each patient to capture the role of geographic variations in practice patterns (contextual factor). We also used binary year indicators to control for intertemporal effects, using 2002 as the reference year.


We preformed bivariate comparisons of the probabilities of having a THR/TKR procedure and other covariates between racial groups, using Whites as the reference group. Chi-squared tests were used for categorical variables, and the corresponding p-values were reported. All regression models used sampling weights provided in NIS to reflect a nationally-representative sample of US hospital inpatient discharges. All statistical analyses were performed using Stata 10 (StataCorp LP, College Station, TX).

Multivariable logistic regression models were employed to estimate the adjusted racial impacts on the probability of actually receiving THR/TKR, respectively. The following models were estimated:

INSERT DESCRIPTION - inline graphic
INSERT DESCRIPTION - inline graphic

where Prob(THR) (or Prob[TKR]) is a binary variable equal to 1 if the patient received THR (or TKR) and 0 otherwise; RACE is a vector indicating patient's race/ethnic groups: [End Page 138] Whites, African-American, or Hispanics; X is a vector of other patient's sociodemographic characteristics, and factors impacting physician decisions, such as admission sources; Y is a vector of hospital characteristics; Z is a variable measuring the area contextual factor; α0, α1, β, Θ, and γ are coefficients to be estimated; and ε is the error term.

We then employed the Blinder-Oaxaca decomposition methods to parse out the most important factors associated with racial and ethnic disparities in the probability of having THR/TKR procedures. This decomposition method has been used extensively to assess mean outcome differences in the labor economics literature.30 -32 In health services research, this method has been employed to study racial and ethnic disparities in different measures of health care access and utilization and health insurance coverage. The Blinder-Oaxaca approach is a regression-based method. To decompose the disparities in THR (or TKR) between Whites and Hispanics,33 -34 we estimated multivariable logistic regressions for these two groups separately. We then subtracted these two estimated equations and decomposed the differences into two parts: 1) differences due to all of the observed population characteristics, (i.e., all of the control variables), and 2) differences reflecting unobserved heterogeneity, such as physician-patient relationships, health preferences and cultural factors.


Descriptive statistics

Whites had a significantly higher probability of receiving total joint replacement procedures (0.96 for THR and 0.95 for TKR) compared with Hispanics (0.92 for THR and 0.89 for TKR) and African Americans (0.92 for THR and 0.89 for TKR). The average age of Whites undergoing these procedures (age=67 for THR and age=68 for TKR) was also higher than that of African Americans (age=63 for THR and age=64 for TKR) and Hispanics (age=64 for THR and age=66 for TKR). Whites were more likely to have a higher median household income than African Americans and Hispanics. Approximately 8% of African Americans and 10% of Hispanics had Medicaid as their primary payer, compared with just 1% of Whites. Only 2% of Whites were admitted through the emergency department, compared with 7% of African Americans and 4% of Hispanics. Most hospital admissions were elective. Among the racial groups, African Americans were least likely to have elective admissions (0.85 compared with 0.92 of Whites and 0.91 of Hispanics for THR; 0.84 compared with 0.91 of Whites and 0.89 of Hispanics for TKR).

Multivariable regression results

After controlling for all of the covariates, African Americans remained significantly less likely to have THR and TKR procedures compared with Whites. The odds ratio of African Americans for THR was 0.57 (p<.001) and for TKR was 0.76 (p<.001). The odds ratio of Hispanics for THR was 0.49 (p<.001). The odds ratio of TKR was 0.66 for Hispanics, and this effect approached statistical significance (p=.07).

Compared with patients age 45 to 54, people under 45 years old and above 85 years old were much less likely to receive these procedures. Females were less likely to receive THR (OR=0.85, p<.001), but more likely to receive TKR (OR=1.07, p<.01). People with higher median incomes and private health insurance were more likely to receive these procedures. Patients admitted through the hospital emergency department were [End Page 139] 97% less likely to receive THR or TKR surgery (OR=0.03, p<.001 for both). Elective admissions were three to four times more likely to receive surgeries. Patients admitted by hospitals with large bed sizes were 89% and 72% more likely to receive THR/TKR, respectively, compared to those admitted by hospitals with a small bed size.

Decomposition results

Total hip replacement. The predicted probabilities of receiving hip replacement procedures were highest for Whites (0.96), followed by Hispanics (0.92) and African Americans (0.92). Observed differences in population characteristics explained 67% and 55% of the differences between Whites vs. African American and Hispanics, respectively.

Among the individual factors, hospital admission source was the most important factor associated with these disparities. It alone explained 77% and 57% of the total observed differences between Whites vs. African Americans and Hispanics, respectively. Compared with Whites, 5% more African Americans and 2% more Hispanics were admitted through the emergency department (ED) (Table 2). However, patients admitted from the ED were far less likely to receive either replacement procedure compared to patients admitted through routine care. Our results showed that if there were no differences in admission sources between Whites and African Americans and Hispanics, the disparities in receiving the procedures would decline by 77% and 57%, respectively, with other factors controlled.

In addition, admission type (elective or not) explained 31% of the disparities between Whites and African Americans. Differences in median area family income explained 15% of the total disparities between Whites and African Americans. Hospital bed size and diagnosis codes were important in explaining differences between Whites and Hispanics. The negative effects of age indicated that without age differences, minorities would be less likely to receive these procedures, and the disparities would increase.

The results also revealed regional and yearly impacts on disparities between Whites vs. Hispanics. Unobserved heterogeneity explained 33% and 45% the disparities between Whites vs. African Americans, and Whites vs. Hispanics, respectively.

Total knee replacement. Similar trends were apparent with respect to racial disparities in TKR. Observed population and hospital characteristics explained 78% of the disparities between Whites vs. African American and Hispanics. Hospital admission source explained 50% of the total explained differences between Whites and African Americans and 26% of the observed differences between Whites and Hispanics. Hospital admission type (elective or not) explained 14% of the disparities between Whites and African Americans. Hospital ownership explained 14% of the White vs. Hispanic disparities. Specific ICD9 codes explained approximately 46% of the TKR disparities for both African Americans and Hispanics (Tables 3 and 4).


Arthritis is one of the most prevalent chronic diseases, affecting millions of patients in the United States.2-4 Total hip replacement and TKR can significantly improve patients' quality of life and reduce their work limitations.5-7 While racial disparities in receiving THR/TKR have been documented, evidence of factors associated with these disparities is limited and inconsistent. By employing the Blinder-Oaxaca decomposition method, [End Page 140]

Variable Definitions and Sample Statisticsa
Click for larger view
Table 2. 

Variable Definitions and Sample Statisticsa

[End Page 141]

[End Page 142]

[End Page 143]

[End Page 144]

our study not only estimated how much of the racial disparities might be explained by observed factors versus unobserved heterogeneity, but also quantified the importance of specific factors in accounting for these differences.

Our results showed that the patients' hospital admission source was one of the most important factors associated with the disparities between Whites vs. African Americans and Hispanics. This result is consistent with previous studies showing that racial minorities are less likely to be admitted via physician referrals,35 -37 more likely to experience treatment delays,38 -39 and more likely to be admitted through the emergency department.21 Previous empirical evidence also indicated that earlier treatment initiation had important health benefits, including improved outcomes for mental health problems and chronic medical conditions such as arthritis.40 -44 Failing to obtain timely treatment may lead to more severe illness and unnecessary medical costs, increasing the economic burden to the health care system. Educational programs may prove beneficial in helping minorities to better understand the value and availability of joint replacement procedures. Programs such as health education, health information and risk reduction activities, have been considered important strategies to promote better general health for under-represented minorities.

Our results revealed that clinical need as measured by ICD9 codes was also important in explaining the disparities in receiving TKR procedures among Whites vs. African American and Hispanics, and the disparities in receiving THR procedures among Whites vs. Hispanics. For example, summary statistics showed that among our samples, African Americans and Hispanics were more than twice as likely as to be diagnosed with 714.00 than Whites. Compared with ICD9 715, patients with ICD9 714.00 codes were only 1% as likely to have the procedure. Thus, the racial disparities in receiving the procedure could be largely contributed to the African Americans' and Hispanics' lower clinical needs to receive the procedure. (That is, there is an argument that minorities' lower utilization might reflect lower clinical needs.) However, most studies of racial disparities in THR/TKR either did not measure clinical need, or only relied on self-reports of clinical needs, such as walking limitations, joint pains, or stiffness.10 Few studies used objective diagnoses such as ICD9 codes.23 In contrast, we employed 5-digit ICD9 codes to capture the clinical appropriateness of the procedures, and found that these objective measures of clinical need also contributed to explaining racial disparities. We recommended that future study on racial and ethnic disparities use specific diagnostic codes in order to control for patients' clinical needs better.

Previous research showed that racial disparities persisted "despite lack of difference in pain perception and joint functionality and higher prevalence rates for osteoarthritis among minority patients than Whites."11 [p.618] After controlling for hospital admission source, hospital and patient characteristics, we also found that disparities persisted. Unobserved heterogeneity, such as physician-patient relationships, health preferences and cultural factors, might explain some of these disparities. Previous work also showed that minorities often have worse physician-patient relationships than Whites.45 -46 This, too, might help explain minorities' lower THR/TKR rates. Suarez-Almazor found evidence suggesting that minorities had different preferences for receiving surgeries.47 Compared with Whites, minorities might be less willing to agree to have the THR/TKR [End Page 145]

Multivariable Logistic Regressions for the Probability of Receiving the THR/TKR Proceduresa
Click for larger view
Table 3. 

Multivariable Logistic Regressions for the Probability of Receiving the THR/TKR Proceduresa

[End Page 146]

procedures. Future research should recognize that factors that might not be easily observable may nonetheless exert powerful effects on the utilization of THR/TKR.

These results should be considered in light of some important limitations. First, we used hospital discharge level data.11 We did not have individual identifiers in the data set. Thus, some individuals might be counted twice if they have TKR surgeries on each knee at two separate times in the same year (staged bilateral knee replacement), but

Decomposition Results of Racial Disparities in Receiving the THR/TKR Proceduresa
Click for larger view
Table 4. 

Decomposition Results of Racial Disparities in Receiving the THR/TKR Proceduresa

[End Page 147]

the rate of the staged bilateral total knee replacements is only 2% of all the TKR procedures.48 Second, due to the lack of individual identifiers, we were unable to determine whether patients who did not receive THR/TKR procedures at the ED might return for another hospital admission for the procedures. Third, our results showed higher surgery rates than are shown in the existing literature. This occurred because we confined our analysis to patients who were most eligible to have the appropriate surgeries. Fourth, HCUP only provided information on patients' inpatient care of community hospitals. Thus, we did not observe outpatient THR/TKR surgeries or THR/TKR surgeries conducted in other types of hospitals, such as long term general hospitals. Finally, while we controlled for a variety of factors, our models did not include some factors that collectively account for a significant portion of racial and ethnic disparities in THR/TKR. As noted above, racial minorities' lower likelihood of receiving THR/TKR surgeries may reflect their less favorable physician-patient relationships. Recent research also suggests that racial and ethnic minorities' cultural backgrounds may be associated with their lower likelihood of receiving THR/TKR and a greater likelihood of failing to receive these procedures in a timely and appropriate fashion.47 For example, minorities may be less familiar with THR/TKR procedures, and they may perceive the benefits of the surgeries differently compared from their White counterparts. Since HCUP is an administrative database, we were unable to capture these unobserved factors, such as cultural norms, as explanatory factors that might be related to the rates of the THR/TKR procedures. Accounting for the role of these unobserved factors is an important direction for further study when appropriate data become available.


Substantive racial and ethnic disparities exist in utilization of THR and TKR surgery. Observed population characteristics accounted for most of these differences, with hospital admission source being the key factor. Unobserved heterogeneity, which might reflect physician-patient relationships, health preferences and cultural factors, were also related to these disparities. Our results highlighted the importance of improving health care access for racial minority patients to ensure they can receive timely joint replacement surgery. In addition, educational programs should be implemented to increase understanding of the value and availability of joint replacement surgery among minorities.

Jie Chen, John A. Rizzo, Shreekant Parasuraman and Candace Gunnarsson  

The authors are affiliated with the Department of Health Services Administration, School of Public Health, University of Maryland at College Park (JC), the Departments of Economics and Preventive Medicine at Stony Brook University (JAR), Depuy Orthopedics in Warsaw, Indiana (SP), and S2 Statistical Solutions, Inc. in Cincinnati (CG).

Please address correspondence to Dr. Chen at Department of Health Services Administration, School of Public Health, University of Maryland; 3310A School of Public Health Building, College Park, MD 20742; (301) 405-9053; jichen@umd.edu.

APPENDIX—RACE Variable was not Available in the Following 10 States:

  • GA   Georgia Hospital Association

  • IL   Illinois Department of Public Health

  • KY   Kentucky Cabinet for Health and Family Services

  • ME   Maine Health Data Organization

  • MN   Minnesota Hospital Association

  • NV   Nevada Department of Health and Human Services

  • OH   Ohio Hospital Association

  • OR   Oregon Association of Hospitals and Health Systems

  • WA   Washington State Department of Health

  • WV   West Virginia Health Care Authority [End Page 148]


1. Centers for Disease Control and Prevention. Prevalence of doctor-diagnosed arthritis—attributable activity limitation—United States, 2003-2005. MMWR Morb Mortal Wkly Rep. 2006; 55(40):1089-92.

2. Centers for Disease Control and Prevention. Prevalence of disabilities and associated health conditions among adults—United States, 1999. MMWR Morb Mortal Wkly Rep. 2001 Feb; 50(7):120-5.

3. Stoddard S, Jans L, Ripple J, et al. Chartbook on work and disability in the United States, 1998: an info use report. U.S. National Institute on Disability and Rehabilitation Research, 1998. Available at: http://www.infouse.com/disabilitydata/workdisability/3_2.php.

4. Hootman JM, Helmick CG. Projections of US prevalence of arthritis and associated activity limitations. Arthritis Rheum. 2006 Jan;54(1):226-9.

5. Ibrahim SA. Racial variations in the utilization of knee and hip joint replacement: an introduction and review of the most recent literature. Curr Orthop Pract. 2010 Mar; 21(2):126-31.

6. Chang RW, Pellissier JM, Hazen GB. A cost-effectiveness analysis of total hip arthroplasty for osteoarthritis of the hip. JAMA. 1996 Mar; 275(11):858-65.

7. Emejuaiwe N, Jones A, Ibrahim S, et al. Disparities in joint replacement utilization: a quality of care issue. Clin Exp Rheumatol. 2007 Nov-Dec; 25(6 Suppl 47):44-9.

8. Skinner J, Weinstein JN, Sporer SM, et al. Racial, ethnic, and geographic disparities in rates of knee arthroplasty among medicare patients. N Engl J Med. 2003 Oct; 349(14):1350-9.

9. Dunlop DD, Manheim LM, Song J, et al. Age and racial/ethnic disparities in arthritis-related hip and knee surgeries. Med Care. 2008 Feb; 46(2):200-8.

10. Steel N, Clark A, Lang IA, et al. Racial disparities in receipt of hip and knee joint replacements are not explained by need: the health and retirement study 1998-2004. J Gerontol A Biol Sci Med Sci. 2008 Jun; 63(6):629-34.

11. Basu J, Mobley LR. Trends in racial disparities among the elderly for selected procedures. Med Care Res Rev. 2008 Oct;65(5):617-37.

12. Ang DC, Ibrahim SA, Burant CJ, et al. Is there a difference in the perception of symptoms between African Americans and whites with osteoarthritis? J Rheumatol. 2003 Jun;30(6):1305-10.

13. Murphy L, Schwartz TA, Helmick CG, et al. Lifetime risk of symptomatic knee osteoarthritis. Arthritis Rheum. 2008 Sep;59(9):1207-13.

14. Hanchate AD, Zhang Y, Felson D, et al. Exploring the determinants of racial and ethnic disparities in total knee arthroplasty: health insurance, income, and assets. Med Care. 2008 May;46(5):481-8.

15. Andersen R, Newman JF. Societal and individual determinants of medical care utilization in the United States. Milbank Mem Fund Q Health Soc. 1973 Winter;51(1):95-124.

16. Cook BL, Manning WG. Measuring racial/ethnic disparities across the distribution of health care expenditures. Health Serv Res. 2009 Oct; 44(5 Pt 1):1603-21.

17. McGuire TG, Alegria M, Cook BL, et al. Implementing the Institute of Medicine definition of disparities: an application to mental health care. Health Serv Res. 2006 Oct; 41(5):1979-2005.

18. Smedley BD, Stith AY, Nelson AR, (eds.). Unequal treatment: confronting racial and ethnic disparities in health care. Washington, DC: National Academies Press, 2003. [End Page 149]

19. Ullman R, Block JA, Stratmann WC. An emergency room's patients: their characteristics and utilization of hospital services. Med Care. 1975 Dec; 13(12):1011-20.

20. Kum-Nji P, Mangrem CL, Wells PJ, et al. Black/white differential use of health services by young children in a rural Mississippi community. South Med J. 2006 Sep; 99(9):957-62.

21. Khaliq AA, Broyles RW. Hospital admissions: who Is admitted through the emergency department? Health Serv Manage Res. 2006 Feb;19(1):13-22.

22. Agency for Healthcare Research and Quality. Healthcare cost and utilization project: a Federal-State-Industry partnership in health data. Rockville, MD: Agency for Healthcare Research and Quality, 2009. Available at: http://www.hcup-us.ahrq.gov/db/nation/nis/NIS_2009_INTRODUCTION.pdf.

23. Mehrotra C, Remington PL, Naimi TS. Trends in total knee replacement surgeries and implications for public health,1990-2000. Public Health Rep. 2005 May-Jun; 120(3): 278-82.

24. National Institute of Health/NIH Consensus Conference. Total hip replacement: NIH consensus development panel on total hip replacement. JAMA. 1995 Jun; 273(24): 1950-6.

25. Palmer SH, Cross M. Total knee arthroplasty. New York, NY: Medscape Reference, 2009. Available at: http://emedicine.medscape.com/article/1250275-overview.

26. Centers for Disease Control and Prevention. Racial disparities in total knee replacement among medicare enrollees—United States, 2000-2006. MMWR Morb Mortal Wkly Rep. 2009 Feb; 58(6):133-8.

27. Kressin NR, Petersen LA. Racial differences in the use of invasive cardiovascular procedures: review of the literature and prescription for future research. Ann Intern Med. 2001 Sep; 135(5):352-66.

28. Phillips KA, Morrison KR, Andersen R, et al. Understanding the context of healthcare utilization: assessing environmental and provider-related variables in the behavioral model of utilization. Health Serv Res. 1998 Aug; 33(3 Pt 1):571-96.

29. Losina E, Wright EA, Kessler CL, et al. Neighborhoods matter: use of hospitals with worse outcomes following total knee replacement by patients from vulnerable populations. Arch Intern Med. 2007 Jan;167(2):182-7.

30. Blinder AS. Wage discrimination: reduced form and structural estimates. J Human Resour. 1973 Autumn;8(4):436-55.

31. Oaxaca R. Male-female wage differentials in urban labor markets. Int Econ Rev. 1973 Oct;14(3): 693-709.

32. Jann B. The Blinder-Oaxaca decomposition for linear regression models. College Station, TX: The Stata Journal. 2008; 8(4):453-79.

33. Fairlie R. 2005. An Extension of The Blinder-Oaxaca decomposition technique to logit and probit models. Journal of Economic and Social Measurement. 2005;10(30):305-16.

34. Bartus T. Marginal effects and extending the Blinder-Oaxaca decomposition to nonlinear models. London, UK: Presentation at the 12th UK Stata Users Group meeting, 2006 Sep 11-12. Available at:http://repec.org/usug2006/bartus_2006_london_p1.pdf.

35. Cooper-Patrick L, Gallo JJ, Gonzales JJ, et al. Race, gender, and partnership in the patient physician relationship. JAMA. 1999 Aug; 282(6):583-9.

36. Ibrahim SA, Whittle J, Bean-Mayberry B, et al. Racial/ethnic variations in physician recommendations for cardiac revascularization. Am J Public Health. 2003 Oct; 93(10): 1689-93.

37. Schulman KA, Berlin JA, HarlessW, et al. The effect of race and sex on physicians' recommendations for cardiac catheterization. N Engl J Med. 1999 Feb; 340(8):618-26. [End Page 150]

38. DiMatteo MR. Variations in patients' adherence to medical recommendations: a quantitative review of 50 Years of research. Med Care. 2004 Mar;42(3):200-9.

39. Vermerie E, Hearnshaw H, Van Royen P, et al. Patient adherence to treatment: three decades of research: a comprehensive review. J Clin Pharm Ther. 2001 Oct; 26(5): 331-42.

40. Simpson SH, Eurich DT, Majumdar S, et al. A meta-analysis of the association between adherence to drug therapy and mortality. BMJ. 2006 Jul;333(7557):15.

41. Wang PS, Berglund P, Olfson M, et al. Failure and delay in initial treatment contact after first onset of mental disorders in the national comorbidity survey replication. Arch Gen Psychiatry. 2005 Jun;62(6):603-13.

42. Sokol MC, McGuigan KA, Verbrugge RR, et al. Impact of medication adherence on hospitalization risk and healthcare cost. Med Care. 2005 Jun; 43(6):521-30.

43. McCombs JS, Thiebaud P, McLaughlin-Miley C, et al. Compliance with drug therapies for the treatment and prevention of osteoporosis. Maturitas. 2004 Jul; 48(3):271-87.

44. Balkrishnan R. The importance of medication adherence in improving chronic-disease related outcomes: what we know and what we need to further know. Med Care. 2005 Jun; 43(6):517-20.

45. Ghods BK, Roter DL, Ford DE, et al. Patient-physician communication in the primary care visits of African Americans and whites with depression. J Gen Intern Med. 2008 May; 23(5):600-6.

46. Johnson RL, Roter D, Powe N, et al. Patient race/ethnicity and quality of patient-physician communication during medical visits. Am J Public Health. 2004 Dec; 94(12): 2084-90.

47. Suarez-Almazor ME, Souchek J, Kelly PA, et al. Ethnic variation in knee replacement-patient preferences or uninformed disparity? Arch Inter Med. 2005 May;165(10):1117-24.

48. Ritter MA, Harty LD, Davis KE, et al. Simultaneous bilateral, staged bilateral, and unilateral total knee arthroplasty: a survival analysis. J Bone Joint Surg Am. 2003 Aug; 85-A(8):1532-7. [End Page 151]


In the HCUP-NIS data, not all the hospitals' ownership types are reported. We included all hospitals that did not report ownership type into the category "unreported type."