|Home | About | Journals | Submit | Contact Us | Français|
Prior studies have indicated racial differences in patients’ expectations for joint replacement surgery outcomes. The goal of this study was to measure these differences using a well-validated survey instrument and to determine if the differences could be explained by racial variation in disease severity, socioeconomic factors, literacy, or trust.
Detailed demographic, clinical, psychological, and social data were collected from 909 male patients (450 African American, 459 white) ages 50–79 years with moderate or severe osteoarthritis (OA) of the hip or knee receiving primary care at 2 veterans affairs medical centers. The previously validated Joint Replacement Expectations Survey was used to assess expectations for pain relief, functional improvement, and psychological well-being after joint replacement.
Among knee OA patients (n = 627), the unadjusted mean expectation score (scale 0–76) for African American patients was 48.7 versus 53.6 for white patients (mean difference 4.9, P < 0.001). For hip OA patients (n = 282), the unadjusted mean expectation score (scale 0–72) for African Americans was 45.4 versus 51.5 for whites (mean difference 6.1, P < 0.001). Multivariable adjustment for disease severity, socioeconomic factors, education, social support, literacy, and trust reduced these racial differences to 3.8 points (95% confidence interval [95% CI] 1.2, 6.3) among knee OA patients and 4.2 points (95% CI 0.4, 8.0) among hip patients.
Among potential candidates for joint replacement, African American patients have significantly lower expectations for surgical outcomes than white patients. This difference is not entirely explained by racial differences in demographics, disease severity, education, income, social support, or trust.
Racial variations in the utilization of major medical and surgical procedures and subsequent health care outcomes have been widely documented in the US health care system, including the Veterans Health Administration (VA) (1). Although the Department of Health and Human Services, the Department of Veterans Affairs, and the National Institutes of Health have established the reduction of health care disparities among minorities as a major national health policy goal (2), and many studies have documented the pervasiveness of racial disparities in health care (3), relatively few studies have examined the root causes for racial variation in health care, such as racial differences in expectations of the outcomes of major surgical procedures.
Joint replacement surgery is a reliable and effective form of treatment for patients with moderate to severe osteoarthritis (OA) of the hip or knee. Joint replacement provides substantial pain relief and functional improvement in ~90% of patients undergoing surgery (4). Despite the effectiveness of joint replacement in reducing pain and improving functional status and quality of life, racial and ethnic minority patients with advanced knee OA are half as likely to receive knee replacement compared with white patients (5,6). Earlier studies conducted in the VA reported that, compared with white patients, African American patients with knee or hip OA perceived joint replacement as a less helpful treatment option (7,8). Moreover, a substantial number of minority patients with OA predicted that joint replacement would have no significant effect on their health status (9).
Although these preliminary findings are suggestive of racial differences in attitudes toward joint replacement, the magnitude of the racial differences remains unclear. It is also uncertain if racial differences in expectations of surgical outcomes may be explained by racial variation in modulating factors such as disease severity, socioeconomic factors, education, literacy, social support, or trust in the health care system. Therefore, the goal of the present study was to measure these differences using a well-validated survey instrument and to determine if the differences could be explained by confounders.
The research protocol was approved by the institutional review boards for both the Philadelphia and Pittsburgh VA Medical Centers. A comprehensive cross-sectional survey was designed to measure clinical, psychological, and socioeconomic factors associated with expectations for the outcomes of joint replacement. The survey was targeted specifically toward patients with moderate to severe OA of the knee or hip between the ages of 50 and 79 years, as these patients would generally be considered good candidates for joint replacement surgery.
Patients were enrolled from October 2004 through September 2006 at the Pittsburgh and the Philadelphia VA Medical Centers. Posters and pamphlets describing the study were placed in the waiting rooms of the primary care clinics and other areas of the hospitals, as well as in 3 satellite outpatient clinics in the Pittsburgh VA Health Care System. In addition, letters were sent to randomly selected patients who had upcoming primary care appointments at each of these sites. These announcements invited patients ages 50–79 years who had chronic knee or hip pain on most days for at least 1 month during the past 6 months to enroll in the study. Patients were advised to contact research coordinators either in person (coordinators were situated in the clinic waiting areas) or by phone if they were interested in enrollment. None of the study publicity material specifically indicated that exploration of racial disparities was a primary goal of the study, although this information was not withheld from any participant who inquired.
At the initial meeting with study personnel, potential enrollees were screened for inclusion and exclusion criteria by completing a brief, 3-stage face-to-face or telephone interview. In screening stage 1, patients were excluded from the study if they were not between ages 50 and 79 years, or if they had a history of knee or hip replacement, lower extremity amputation, a medical illness with a life expectancy <12 months, inflammatory arthritis (e.g., rheumatoid arthritis, systemic lupus erythematosus, ankylosing spondylitis, psoriatic arthritis, gout), or dementia. In screening stage 2, the Arthritis Supplement of the First National Health and Nutrition Examination Survey instrument was administered to identify individuals who were highly likely to have a diagnosis of knee or hip OA (10,11). Finally, in screening stage 3, those patients who screened positive for OA in stage 2 were assessed for the clinical severity of their OA using the Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) (12), a reliable and valid scale designed to quantify pain and functional status in patients with OA (13). Summary WOMAC scores range from 0 (no pain or disability) to 100 (severe pain with total disability), and patients with scores ≥39 are likely to experience substantial symptom relief and functional improvement from joint replacement surgery (14). Enrollment in this study was limited to patients with WOMAC scores ≥39 on the screening assessment; these patients proceeded to the full interview. Patients who completed all study procedures were reimbursed $25 for their time and effort.
The full data collection interview consisted of an assessment of patients’ demographic and socioeconomic characteristics, quality of life, pain level, functional status, literacy, trust in physicians, social support, and expectations for joint replacement surgical outcomes. Quality of life was measured using the Medical Outcomes Study (MOS) Short Form 12 (SF-12), version 2 (15). Pain level was measured using a 0–100 visual analog “thermometer.” Functional status was assessed using the WOMAC as described above. Medical literacy was measured using the Rapid Estimate of Adult Literacy in Medicine-Revised (REALM-R) instrument (16). Patients’ trust in their physicians was measured using the Trust In Physician Survey (17,18). The MOS Modified Social Support Survey was used to assess the quality of patients’ social support systems (19). Expectations of the effectiveness of joint replacement surgery in improving quality of life were measured using the Hospital for Special Surgery Joint Replacement Expectations Survey (JRES). This instrument, developed at an orthopedic hospital in New York City, was separately developed by Mancuso et al for hip and knee surgical candidates through an extensive process by which draft surveys containing 52 potential categories of expectations were administered, and the survey elements with the highest kappa statistics (intrarater reliability) were retained in the final instruments, which were subsequently rigorously tested in validation cohorts (20–22). The JRES contains 19 (knee) or 18 (hip) items, with a 5-point Likert-scale response range for each item where a strong expectation of a particular (positive) outcome is scored as 4, while the absence of an expectation for the outcome is scored as 0 (Appendices A and B, available at the Arthritis Care & Research Web site at http://www.interscience.wiley.com/jpages/0004-3591:1/suppmat/index.html).
Patients self-identified their race and ethnicity, and only the responses of patients who self-identified as African American or white were included in this report. Additionally, because fewer than 5% of our respondents were women, we excluded these responses from our analyses.
We conducted comparative analyses comparing African American and white patients using chi-square tests for categorical variables, t-tests for normally distributed continuous variables, and Wilcoxon’s rank sum tests for non-normally distributed variables. Survey responses were subsequently analyzed using multivariable regression models. These models were fitted as an ordinary least squares regression with the JRES score as the dependent variable. Independent variables in this model included race, age, sex, categorized income, categorized education, employment status, marital status, pain score, functional status (WOMAC score), location of interview (Pittsburgh or Philadelphia), literacy (REALM-R), social support, and trust in physicians. We then added interaction effects between race and other key predictors to our fully specified model to determine if the effect of these predictors was modulated by race. Finally, we fit the fully specified model (excepting race) separately to white and African American patients to assess for any large differences in the effects of predictors.
All statistical analyses were performed using SAS 9.1 (SAS Institute, Cary, NC). All significance tests were 2-sided, with a P value less than 0.05 assumed to be significant.
The final survey cohort comprised 909 primary care patients with chronic knee or hip OA who identified themselves as either white or African American. Characteristics of white (n = 459) and African American (n = 450) enrollees are summarized in Table 1. African American patients were younger, predominantly from Philadelphia, less likely to be married, more likely to have household incomes below the poverty level, and less likely to have received an associate’s degree or other higher education degree than white enrollees. More African American respondents (42%) than white respondents (32%) reported receiving disability payments due to their knee/hip OA (P < 0.001).
Mean survey instrument scores are summarized by race in Table 2. African American patients had lower health literacy and lower social support than whites, although there was no racial difference in trust in physicians. African American and white patients had similar Physical Component Summary scores on the SF-12, although African American patients had significantly lower Mental Component Summary scores (P = 0.01). African American patients had significantly higher mean pain scores as reported on a visual analog scale (P < 0.001), and African American patients had significantly more functional impairment from their OA as measured by the WOMAC index (P < 0.001).
To assess reliability of the JRES instrument in our patient population, we calculated Cronbach’s alpha for the entire cohort of knee OA respondents (α = 0.96) and hip OA respondents (α = 0.95). Because the JRES was primarily developed and validated among white patients (20,21), we then calculated Cronbach’s alpha separately for African American knee OA patients (α = 0.95) and white knee OA patients (α = 0.96), as well as African American hip OA patients (α = 0.93) and white hip OA patients (α = 0.96). All alpha values exceeded 0.9, indicating extremely high reliability of the scale among both African Americans and whites.
Among patients with knee OA, white patients had a mean ± SD unadjusted JRES (expectation) score of 53.6 ± 14.5, whereas African American patients had a mean ± SD score of 48.7 ± 14.6 (P < 0.001 for the difference). More African American patients (24%) than white patients (15%; P = 0.004) had JRES scores below the midpoint of the scale, indicating predominantly low expectations for the outcomes of surgery. JRES scores for white and African American patients with knee OA are shown in Figure 1.
Among hip OA patients, white patients had a mean ± SD unadjusted JRES score of 51.5 ± 14.3, whereas African American patients had a mean ± SD score of 45.4 ± 14.2 (P < 0.001 for the difference). Approximately 25% of African American patients, compared with 16% of white patients (P = 0.06), had JRES scores below the scale midpoint. JRES scores for white and African American patients with hip OA are shown in Figure 2.
In fitting the ordinary least squares multivariable regression models, we performed Kolmogorov-Smirnov tests for normality of the regression residuals for the fully specified models. These tests failed to reject the null (P > 0.05 in all cases), indicating that the models met the normality assumption for ordinary least squares regression.
Results from multivariable ordinary least squares regression models are summarized in Table 3. For knee OA patients, adjustment for WOMAC score, age, sex, marital status, and clinical site (Philadelphia or Pittsburgh) reduced the mean racial difference in knee JRES scores from an unadjusted difference of 4.9 points (95% confidence interval [95% CI] 2.6, 7.2) to a difference of 4.0 points (95% CI 1.5, 6.6). Addition of employment status, income, and education to the model further reduced the race difference to 3.7 points (95% CI 1.2, 6.3). Addition of the REALM-R literacy score, Trust in Physician score, and MOS Social Support score did not materially influence the racial difference in mean scores, which was 3.8 points (95% CI 1.2, 6.3).
For hip OA patients (Table 4), adjustment for the WOMAC score, age, sex, marital status, and clinical site (Philadelphia or Pittsburgh) reduced the mean racial difference in hip JRES scores from an unadjusted mean difference of 6.1 points (95% CI 2.8, 9.5) to 4.2 points (95% CI 0.4, 8.1). Addition of employment status, income, and education to the model did not materially change the racial difference in means scores, which was 4.4 points (95% CI 0.6, 8.2). Addition of the REALM-R literacy score, trust score, and social support score also had a minimal effect on the race difference in mean scores, which was 4.2 points (95% CI 0.4, 8.0).
To assess whether key predictor variables interacted with race, we then fit our full regression model including interaction terms between race and marital status, employment, education, income, trust in physicians, and social support. None of the race-interaction coefficients were significant (P > 0.1 for all) for either knee or hip OA patients. To further test for differences in the effects of predictors by race, we next fit the full regression models (excepting the race variable) separately for white patients and African American patients. For knee OA patients, a single-point increase in the Trust in Physicians scale was associated with a 0.46 higher JRES score (P = 0.04) for whites, and a 0.41 higher JRES score (P = 0.001) for African Americans. Higher WOMAC scores were associated with a lower JRES score for white knee OA patients (−0.16 per unit increase in WOMAC; P = 0.04) but not for African American knee OA patients (0.04 per unit increase in WOMAC; P = 0.51). For white hip OA patients, increased trust in physicians was associated with a 0.31 higher JRES score (P = 0.08), and this effect was slightly more pronounced for African Americans (0.66 higher score; P = 0.002). White hip OA patients who were employed had higher adjusted JRES scores (4.6 higher; P = 0.05), as did employed African American hip OA patients (5.2 higher; P = 0.03). No other covariates were significant predictors of JRES score for either racial group among hip OA patients.
In this sample of primary care patients with chronic knee/hip pain who were potential candidates for joint replacement, we found significant racial differences in expectations for joint replacement surgery outcomes using a comprehensive, well-validated survey instrument. These differences were not explained by racial variation in the age or sex distribution of the patient populations, nor were they explained by differences in geographic location, socioeconomic status, educational level, health literacy, trust in physicians, or social support between African American and white patients.
The clinical or policy significance of a modest, 4–6-point difference in JRES scores, equivalent to a standardized difference (i.e., fraction of a standard deviation) of 0.3–0.4, is uncertain. Because our study was a cross-sectional design, it was not possible to ascertain if patients with lower JRES scores were actually less likely to undergo surgery in subsequent months and years. However, a growing body of evidence suggests a strong correlation between patients’ expectations of joint replacement surgery and their stated willingness to undergo surgery (9,23,24). Future longitudinal studies are necessary to clearly quantify the relationship between patients’ expectations and joint replacement surgery rates.
Our findings add to substantial research that has explored the root causes for racial differences in the utilization of joint replacement surgery for knee and hip OA. Ibrahim et al previously reported that, compared with white patients, African American patients with knee or hip OA are less likely to perceive joint replacement as an effective treatment option (8). Another study by Ibrahim et al assessing patient familiarity with joint replacement and expectations of surgical outcomes found that African American patients were less familiar with the details of joint replacement surgery and the duration of the expected recovery period (23). The current study expands on these previous findings by using a more comprehensive (i.e., 17- or 18-item), integrated (i.e., single composite score), well-validated measure of joint replacement surgery expectations. We also enrolled sufficient numbers of patients to verify these findings separately among patients with knee OA and those with hip OA. Additionally, in multivariable analyses, we adjusted for several factors likely to confound the relationship between race and joint replacement expectations, including other demographic factors, socioeconomic variables, literacy, social support, and trust. Indeed, African American and white patients in our sample differed markedly in age, geographic location, socioeconomic status, educational attainment, health literacy, and social support. These factors are all likely to influence patient expectations of surgery independent of race, yet when we adjusted for the independent effect of these variables, racial differences in patient expectations persisted. Additionally, we found little evidence of interaction between race and other predictors of expectations. Generally, significant predictors of higher expectations, such as employment and greater trust in physicians, were similarly influential in both racial groups.
If racial differences in joint replacement expectations are not mediated by the many factors that we explored, what are the central causes for these racial differences? Among the remaining possibilities are racial differences in the composition of patients’ social networks, through which trusted information regarding the benefit of knee and hip replacement may be obtained, as well as differences in access to other sources of information about joint replacement such as the Internet. Our own prior research suggests that information received through informal channels (rather than, for example, a physician visit) is often critical in influencing patients’ attitudes toward medical technology (25).
With any elective surgery, it is vital that patient-physician decision making truly be shared, and that patient autonomy be respected (26). As such, a decision to decline elective joint replacement surgery by a patient with moderate or severe arthritis is not necessarily irrational, and may reflect an accurate self-assessment of individual circumstances and/or risk preferences (27). However, a decision to decline beneficial therapy based on erroneous, incomplete, or outdated information is not an issue of autonomy or preference, but of disparity in knowledge (9). More African American patients may lack access to accurate information (e.g., interpersonal networks where receipt of joint replacement is common, etc.) about the experiences after surgery than do white patients. In a context in which trust in the health system is low, African American patients may therefore discount the benefits of surgery as described by health care providers (28).
Interventions designed to improve the accuracy of, and reduce racial differences in, patients’ understanding of the outcomes of joint replacement surgery may be a particularly effective way of reducing racial disparities in utilization of this effective treatment option. To be effective, such interventions should not only increase the quality and quantity of information available to patients, but also increase the number and variety of sources of trusted information. Weng et al have pilot tested an educational intervention (a video and tailored decision aid) designed to reduce disparities in expectations of knee replacement, measured by the anticipated WOMAC score after surgery (29). This intervention lowered the expected WOMAC score for both pain and physical function among African American participants. Although these findings are preliminary, they suggest that racial differences in expectations are modifiable. Peer counselors, testimonials, compelling presentations of race- and ethnicity-specific outcome data, culturally competent media presentations, and the establishment of patient networks all may be effective in reducing racial differences in expectations and consequently improve equity in joint replacement surgery rates.
In interpreting our results, it is appropriate to acknowledge several limitations. First, this research is a cross-sectional study, which limits causal inference. In particular, we did not assess whether patients were anticipating surgery in the near future, nor did we later determine who had undergone surgery subsequent to the survey. Respondents who were facing the tangible prospect of joint replacement surgery may have responded differently than those patients for whom surgery was a more hypothetical option. Second, although our recruitment of nearly 1,000 patients over 2 years’ time improved our likelihood of attracting somewhat more reluctant participants to enter the research cohort, it is possible that our method of enrollment selected patients who were generally more trusting of the health system or of the VA in particular. Patients with frequent visits to the medical centers also had more opportunities to be enrolled, so we may have inadvertently enrolled a cohort of patients who were more connected to the health system than typical veterans. Third, this study excluded patients who previously had undergone joint replacement surgery. If one racial group had greater access to surgery than the other regardless of expectations, it is possible that the racial differences we observed were an artifact created by the pool of presurgical knee/hip OA patients in one racial group being depleted faster than the other racial group. However, we believe this phenomenon likely would have diminished the apparent racial difference in expectations, because it is probable that white patients would have had more facilitated access to surgery. As such, it would have been less likely for us to have enrolled white patients who were both highly favorable of joint replacement surgery but who had not yet undergone surgery, thus potentially reducing the mean expectations score among white participants compared with African American participants. Finally, the racial difference we observed could have been an artifact produced by the JRES instrument itself, if the instrument performed differently across racial groups. However, the highly similar distributional form of the responses, the nearly identical Cronbach’s alpha values, and the similar results of the regression analyses stratified by race suggest this is unlikely.
In this study of primary care patients with chronic knee/hip pain who were potential candidates for joint replacement, we found significant racial differences in patient expectations regarding joint replacement surgery. These findings were independently observed for both hip OA and knee OA patients. The racial differences in expectations persisted despite adjustment for multiple potential confounders. Future studies should explore interventions designed to reduce racial disparities in expectations of the benefits and risks of joint replacement. Such interventions are likely to improve the quality of patient decision making, while potentially reducing racial differences in utilization of joint replacement and subsequent quality of life for patients with OA.
The authors gratefully acknowledge the assistance in manuscript preparation provided by Alexis Greenhut, MPH, statistical advice provided by Samuel Field, PhD, and graphical expertise provided by Janell Olah, MFA.
Supported by a Merit Review grant (HSR&D IIR 03-201-1) from the VA Health Services Research and Development Service (Washington, DC). Drs. Groeneveld and Ibrahim’s work was supported by Career Development awards from the VA Health Services Research and Development Service (Washington, DC). Dr. Ibrahim’s work was supported by the National Institute of Arthritis and Musculoskeletal and Skin Diseases (grant K24-AR0-55259).
AUTHOR CONTRIBUTIONSDr. Groeneveld had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study design. Groeneveld, Kwoh, Ibrahim.
Acquisition of data. Groeneveld, Kwoh, Appelt, Gutierrez, Wessel, Ibrahim.
Analysis and interpretation of data. Groeneveld, Kwoh, Mor, Ibrahim.
Manuscript preparation. Groeneveld, Kwoh, Mor, Appelt, Gutierrez, Wessel, Ibrahim.
Statistical analysis. Groeneveld, Mor, Geng.