|Home | About | Journals | Submit | Contact Us | Français|
This study compared pain and function among African Americans and Whites with radiographic hip and/or knee osteoarthritis (OA), controlling for radiographic severity and other patient characteristics.
Participants were 1,368 individuals (32% African American) from the Johnston County Osteoarthritis Project with only knee OA, only hip OA, and both knee and hip OA. Linear regression models examined racial differences in Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) total scores and pain and function subscales, adjusting for radiographic severity, age, gender, education, body mass index (BMI), depressive symptoms, and WOMAC pain (last variable in models of function).
Among those with only knee OA, African Americans had significantly worse mean WOMAC total scores than Whites (32.8 vs 24.3, p<0.001), and worse pain and function scores (p<0.001). Racial differences in WOMAC total, pain, and function scores persisted when controlling for radiographic severity and demographic factors but were not significant when also controlling for BMI and depressive symptoms. In models of WOMAC function, pain was the most strongly associated variable and substantially reduced the association of race with function. There were no racial differences in WOMAC scores among those with only hip OA or with both knee and hip OA.
Among participants with knee OA, racial differences in pain and function may be explained by BMI and depressive symptoms, and racial differences in function may also be largely influenced by pain. Improving management of weight and depressive symptoms may be key steps toward reducing racial disparities in knee OA symptoms.
Previous studies have identified racial differences in arthritis symptoms, with African Americans reporting greater pain and activity limitation than Caucasians (1-4). However, few studies have focused specifically on osteoarthritis (OA), and among these, findings have been inconsistent regarding racial differences in self-reported pain and function (3, 5, 6). Additionally, most of these studies have not examined other demographic and clinical factors that may play a role in the relationship between race and OA symptoms and functional limitations. One study reported that African American and Caucasian male veterans with knee or hip OA did not differ significantly in self-reported pain and function (as measured by the Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC)) when stratified according to radiographic severity or when controlling for radiographic severity, demographic characteristics, body mass index (BMI), comorbidity, and depression (5). However, this study only included patients with moderate and severe knee or hip pain. Another study of veterans with OA found that African Americans had significantly higher WOMAC scores (indicating more pain and poorer function) compared to Caucasians when controlling for other demographic characteristics, but this study did not examine radiographic severity, nor the independent and joint effects of concomitant knee and hip involvement (3). The generalizability of results from both of these patient-based studies of mostly men may be limited.
The purpose of the current study was to compare pain and function between African American and Caucasian men and women with radiographic knee and /or hip OA in a large, community-based sample and examine whether radiographic severity and other demographic or clinical factors might explain any differences. We were also interested in assessing whether these associations varied according to affected joints (only knee OA, only hip OA, or both knee and hip OA) since some evidence suggests racial differences in OA may not be the same for the hip and knee. Specifically, several studies suggest African Americans have more severe radiographic knee OA than Caucasians, but evidence is mixed regarding whether radiographic hip OA differs significantly according to race (6-11).
The cross-sectional sample was composed of individuals enrolled in the Johnston County Osteoarthritis Project, an ongoing, population-based study of the occurrence of knee and hip OA in a rural, bi-racial population of North Carolina. Details of this study have been reported previously(11, 12). Briefly, this study involved civilian, non-institutionalized adults aged 45 years and older who resided in six townships in Johnston County. At baseline, participants were recruited with over-sampling of African Americans. This analysis selected individuals who had participated in either the first follow-up of the study (n = 1,733) conducted between 1999 and 2004, or had been newly enrolled (n = 1,015) in 2003 or 2004. New individuals were enrolled to enrich the sample for African Americans and younger individuals who were deliberately targeted for inclusion. As such, the newly enrolled participants were younger (mean age 59.3 years and 65.8 years, in newly enrolled and first follow-up sub-samples, respectively) and more likely to be African American (40% vs. 28%). However, associations of race with self-reported pain and physical function were similar between the follow-up and newly enrolled groups, and these groups were combined for the current analyses. These analyses included 1,368 individuals who had radiographic evidence of OA in at least one knee or hip. Because we were interested in examining whether the association of race with pain and function differed according to joint site(s), we divided the sample into 3 analytic groups: only knee OA (N=540), only hip OA (N=470), and both hip and knee OA (N=358). These categories were based on knee and hip radiographs only, and individuals in all analytic groups may have also had OA in other joints.
OA symptoms and function were measured using the WOMAC, a widely-used and well established scale designed to assess pain, stiffness, and function in lower extremity OA(13, 14). Total WOMAC scores can range from 0-96, with higher scores indicating greater overall severity. In addition, we examined the pain (range: 0-20) and function (range: 0-68) subscales separately. Because the stiffness subscale only contains two items (range: 0-8), we did not separately examine this subscale.
In addition to participant race, our analyses included other demographic and clinical characteristics that may be potential confounders of associations of race with pain and / or function, based on previous research (15-18). Participants completed two home interviews and a clinic visit, which included measurement of height without shoes and weight with a balance beam scale, and a radiographic examination of the knees and hips. All participants underwent posterior- anterior (PA) radiography of both knees with weight-bearing using a Synaflex® positioning device. Supine A-P pelvis films were obtained on all men and women 50 years of age and older. All radiographs were read for Kellgren-Lawrence (K-L) score by a single bone and joint radiologist (JBR) without regard to clinical status. Inter-rater and intra-rater reliability of the radiologist were high (weighted kappas = 0.9), as previously reported(11). Radiographic knee and hip OA were defined as a K-L grade of at least 2 in at least 1 knee or hip, respectively. As a measure of radiographic OA severity, we created an indicator of whether participants had moderate or severe OA (defined as K-L grade ≥ 3) in at least one joint in the group(s) of interest (e.g., at least one knee with moderate to severe OA for the knee OA only group). Other participant characteristics examined included age, gender, education (≥ 12 years vs. <12 years), body mass index (BMI; calculated as weight in kilograms/height in meters squared and defined continuously), and depressive symptoms, assessed with the Center for Epidemiologic Studies Depression Scale (CES-D) and defined continuously(19).
We first compared demographic and clinical characteristics according to race, within our three analytic groups. Because CES-D scores were not normally distributed, we calculated medians (and interquartile ranges) and used Wilcoxon tests to compare these variables according to race. Other continuous variables (e.g., age, BMI) were examined using means and t-tests, and categorical variables were examined using proportions and Chi-square tests.
We compared mean WOMAC total and subscale scores according to race, using t-tests, for the three analytic groups. Next, we conducted a series of linear regression models examining the associations of race and other covariates with WOMAC total and subscale scores. For models of WOMAC total, pain, and function scores, Model 1 included only race; Model 2 added the indicator of moderate or severe OA; Model 3 added demographic characteristics (age, gender, and education); Model 4 added BMI and depressive symptoms. For analyses of WOMAC function scores, we also added the WOMAC pain subscale score in Model 5 since previous studies have shown pain is a key predictor of subsequent functional decline in longitudinal studies (20, 21).
While we observed that the residuals were not normally distributed in regression models involving the WOMAC total and subscale scores, linear regression may be robust to this assumption. However, because this assumption was violated and transformation did not improve this situation, we also conducted an additional set of analyses to confirm the results of our linear regression models. Briefly, we categorized WOMAC scores into quartiles and performed a series of proportional and partial proportional odds models, as appropriate, including the same covariates described above for the linear regression models. Results of these analyses were consistent with those for the linear regression models. Therefore we present results of the linear regression models for ease of interpretation. All statistical analyses were performed using SAS PC, Version 9 (SAS Inc., Cary, NC).
The sample of 1,368 with radiographic knee and / or hip OA included 32% African Americans, 65% women, and had a mean age of 67 years. In unadjusted analyses (Table 1), all participant characteristics differed significantly by race among participants with only knee OA and those with both knee and hip OA. Specifically, African Americans had higher proportions of individuals with moderate to severe knee OA, women, and individuals who completed less than 12 years of education. African Americans were younger and had higher BMI scores and worse depressive symptoms. For those with only hip OA, the only statistically significant difference was that African Americans had higher scores of depressive symptoms.
Among those with only knee OA, mean WOMAC total and subscale scores were significantly higher among African Americans than Caucasians (Table 1). In linear regression models of WOMAC total, pain, and function subscale scores (Tables 2--4),4), African American race was associated with significantly higher WOMAC scores in Models 1, 2, and 3. However, when adding BMI and depressive symptoms in Model 4, the associations of African American race with WOMAC total and pain scores were no longer statistically significant. In Model 4, the least squares means for WOMAC total scores for African American and Caucasians were 28.3 and 26.3, respectively; least squares means for WOMAC pain scores for African Americans and Caucasians were 5.9 and 5.4, respectively; least squares means for WOMAC function scores for African Americans and Caucasians were 19.7 and 18.3, respectively. We also conducted exploratory analyses of models that included BMI and depressive symptoms separately (along with race, moderate or severe knee OA, and demographic characteristics). The addition of either BMI or depressive symptoms to the model resulted in a non-significant association of race with both WOMAC total and pain scores (data not shown).
There was a very strong association of WOMAC pain score with WOMAC function score (ß=0.81, p<0.001 in Model 5, Table 4), and the model including pain explained 76% of the variance on WOMAC function scores, compared with 26% of the variance in Model 4 that did not include pain. In Model 5, the least squares means for WOMAC total scores for African American and Caucasians were 19.0 and 19.4, respectively.
Mean WOMAC total and subscale scores did not differ according to race among participants with only hip OA or those with knee and hip OA (Table 2). Among participants with only hip OA, African American race was not significantly associated with WOMAC total scores (ß=0.04, p=0.348), WOMAC pain subscale scores (ß=0.06, p=0.191), or WOMAC function scores (ß=0.04, p=0.380) in Model 1, nor were there any significant racial differences in Models 2-4 (data not shown). Similarly, among participants with both knee and hip OA, African American race was not significantly associated with WOMAC total scores (ß=0.03, p=0.563), WOMAC pain subscale scores (ß=0.05, p=0.391), or WOMAC function scores (ß=0.03, p=0.577) in Model 1, nor were there any significant racial differences in Models 2-5 (data not shown).
Among Johnston County Osteoarthritis Project participants with radiographic knee or hip OA, associations of race with pain and function varied by joint site. African Americans with only knee OA had significantly higher WOMAC total, pain, and function subscale scores than Caucasians, even when controlling for the presence of moderate or severe radiographic OA and demographic factors including also included age, gender, and education level. However, when also controlling for BMI and /or depressive symptoms, these association became non-significant. Racial differences in pain have been observed in a number of other populations, but underlying mechanisms are not well understood (22). It is noteworthy that an objective measure of disease severity (i.e., the presence of moderate or severe radiographic disease) did not explain racial differences in pain. Results of this study suggest that among patients with knee OA, BMI and depressive symptoms may be important factors explaining racial differences in pain, as well as physical function.
In models of WOMAC function scores among patients with only knee OA, the WOMAC pain subscale had a stronger association than any other variable, confirming previous research regarding the important association of pain severity with function (20, 21, 23, 24). The model including pain explained substantially more of the variance in function scores than a model that include all other covariates but not pain. Further, addition of WOMAC pain scores into the model resulted in a substantial decrease in the association of race with WOMAC function scores, and in this model, the least squared mean WOMAC function scores were actually slightly lower for African Americans than Caucasians. Results of this study suggest pain may be a key factor underlying racial differences in function among individuals with knee OA; longitudinal studies would be helpful to more clearly examine these interrelationships.
Our findings regarding racial differences in the only knee OA group differ from a previous study that observed no racial differences in WOMAC scores veterans with OA even in unadjusted analyses (5). There are two likely contributors to these differences. First, the previous study involved a sample with mixed knee and hip OA, and our results indicate racial differences in pain may be more pronounced in only knee OA. Second, the study of veterans with OA was restricted to individuals with moderate to severe pain, perhaps suggesting that racial differences in symptoms may be more pronounced when including patients with a broader spectrum of disease severity. It is also noteworthy that the veteran sample comprised primarily men, whereas our sample was over half women. Further, females had worse pain and function than males in this sample.
In contrast to findings among the group with only knee OA, there were no racial differences in WOMAC total score or subscale scores in the group with only hip OA or the group with both knee and hip OA, even in unadjusted analyses. One possible reason for the lack of racial difference among participants with only hip OA may have been that this group had lower symptom severity and variability than the only knee OA group, as well as less severe radiographic disease. Among those with only hip OA, 7% and 4% had K-L grade 3 or 4 OA in the right and left hips, respectively. In comparison, among the only knee OA group 42% and 41% had K-L grade 3 or 4 OA in the right and left knees, respectively. It is possible that among a sample of individuals with more severe radiographic hip OA, racial differences in pain and function may be more pronounced. Since there were racial differences in pain and function among participants with only knee OA, it is interesting that no racial differences were observed in the group with both knee and hip OA. African Americans in the group with both knee and hip OA actually had lower WOMAC scores than their counterparts with only knee OA, whereas the converse was true among Caucasians (Table 2). Reasons for these patterns are not clear and warrant further investigation.
There are several important strengths to this study. The data come from a large community-based sample with a substantial proportion of African Americans that included both men and women. We observed the full spectrum of radiographic OA severity. Radiographs were completed on both knees and hips, which allowed us to conduct analyses separately according to affected joint groups, including those with both knee and hip OA.
There are also some limitations. While we attempted to include a broad spectrum of potential confounders of the associations of race with pain and function, based on previous research (15-18), there may be other explanatory factors not examined in this study. We examined overall radiographic severity, defined by total K-L grade, but did not assess specific radiographic features. We also did not assess radiographic changes at the patellofemoral joint, which can be associated with pain and disability (25). In other analyses of the Johnston County Osteoarthritis Project cohort, African Americans were not more likely than Caucasians to have isolated patellofemoral joint osteophytes (26). The Johnston County Osteoarthritis Project sample is from a relatively rural North Carolina community, and additional research is needed to examine whether these results generalize to other groups.
In summary, this study found that in a community sample, African Americans and Caucasians with only hip OA or both hip and knee OA did not differ in self-reported pain or function. In unadjusted analyses among participants with only knee OA, African Americans had greater self-reported pain and functional limitations than Caucasians, but these differences were explained by other clinical factors. In particular, pain was an important factor explaining racial differences in function, and BMI and depressive symptoms were important factors explaining racial differences in both pain and function. Therefore improving management of obesity and depressive symptoms may be key avenues for reducing racial disparities in symptom severity among patients with knee OA. Additional research is needed to understand differences in the associations of race with pain and function according to joint site(s) affected by OA.
The findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the Centers for Disease Control and Prevention or the Department of Veterans Affairs. Funding was made possible (in part) by: cooperative agreements S1734 and S3486 from the Centers for Disease Control and Prevention / Association of Schools of Public Health, the NIAMS Multipurpose Arthritis and Musculoskeletal Disease Center grant 5-P60-AR30701, and the NIAMS Multidisciplinary Clinical Research Center grant 5 P60 AR49465-03.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.