Abstract
The objective of this analysis was to assess risk factors for self-reported osteoarthritis (OA) in an ethnically diverse cohort of women. The participants were postmenopausal women aged 50 to 79 (n=146,494) participating in the clinical trial and observational study of the Women’s Health Initiative (WHI). Baseline OA and risk factors were collected from WHI questionnaires. Logistic regression was used to find the association between the risk factors and OA. Risk factor distribution and ethnicity interaction terms were used to assess ethnic differences in OA risk. Forty-four percent of the participants reported OA. Older age (odds ratio (OR)70–79 vs 50–59=2.69, 95% confidence interval (CI)=2.60–2.78) and higher body mass index (BMI) (ORBMI≥40.0 vs <24.9=2.80, 95% CI=2.63–2.99) were found to be the strongest risk factors associated with self-reported OA. The prevalence of obesity (BMI≥30.0) was 57.9% in African Americans, 51.0% in American Indians, 41.9% in Hispanic whites, and 32.9% in non-Hispanic whites. The prevalence of other major OA risk factors was higher in African-American, American-Indian, and Hispanic white women than in non-Hispanic white women. Non-Hispanic white women who were in the extreme obese category (BMI≥40.0λkg/m2) had a 2.80 times (95% CI=2.63, 2.99) greater odds of self-reported OA. The odds were even higher in American-Indian (OR=4.22, 95% CI=1.82, 9.77) and African-American (OR=3.31, 95% CI=2.79, 3.91) women, indicating a significant interactive effect of BMI and ethnicity on odds of OA. In conclusion, OA is a highly prevalent condition in postmenopausal women, and there are differential effects according to ethnicity.
Keywords: self-reported OA, WHI, ethnicity, risk factors
INTRODUCTION
Arthritis and rheumatic conditions are the leading cause of disability in the United States, affecting approximately 46 million adults, with estimated total medical costs of $128 billion annually.[1] Osteoarthritis (OA), the most common arthritic condition, is characterized as “a range of disorders that result in structural and functional failure of synovial joints when the dynamic equilibrium between the breakdown and repair of joint tissues is overwhelmed.”[2] The prevalence of OA has been estimated to be 12% in people 25 and older[3] and increases to almost 68% in people aged 65 and older.[4] It has been recently reported that, along with osteoporosis, OA is a major health problem in postmenopausal women, and the condition is more debilitating in this population.[5] Strong OA risk factors identified by epidemiological studies include age and high body mass index (BMI). Physical activity has been shown to be a strong protective factor,[6–8] and factors such as race or ethnicity, educational attainment, cigarette smoking, hypertension, fasting blood glucose, and alcohol intake have been inconsistently associated.[7]
Although several studies have investigated OA and its risk factors in postmenopausal women, few have assessed these risk factors in a group of women from multiple racial and ethnic backgrounds. Research has shown that there are ethnic differences in arthritis-related outcomes, such as disability and pain,[9] and African Americans have an overall higher age-adjusted death rate from arthritis and other rheumatic diseases.[10] The Women’s Health Initiative (WHI) is one of the largest, most ethnically diverse cohorts of postmenopausal women with abundant health information on variables related to OA. The WHI cohort provides a potential resource to shed new light on ethnic variations in the prevalence of and risk factors for OA in postmenopausal women. The objectives of the current analysis were to assess the prevalence of self-reported OA within the WHI, to investigate the association between established OA risk factors and self-reported OA in this population, and to evaluate ethnic differences in the distribution and effect of risk factors within the WHI.
METHODS
Study Population
The focus of the WHI was to investigate the risk factors and preventive strategies of the major contributors to morbidity and mortality in postmenopausal women: heart disease, breast and colorectal cancer, and osteoporotic fractures.[11] The WHI recruited 161,809 postmenopausal women aged 50 to 79 from 40 centers across the country to participate in the clinical trial or observational study. Details of recruitment strategies and baseline participant information have been previously published.[12]
Outcome
This analysis was conducted using baseline self-reported OA from the WHI observational study (n=92,971) and clinical trial (n=68,838). The participants were asked, “Did your doctor ever tell you that you have arthritis?” with choices including yes or no. Nonresponders (n=1,438) were excluded. The participants answering no were placed in the non-OA reference group (n=83,954). Women responding yes (n=76,417) were then asked “What type of arthritis do you have?” including choices of rheumatoid arthritis and other or don’t know. Women reporting rheumatoid arthritis (n=7,862) and women not answering the follow-up arthritis question (n=3,995) were excluded from the analysis. Women selecting other or don’t know were placed in the OA case group (n=64,550). Prevalence of other arthritic and rheumatic conditions (such as systemic lupus erythematosus and Crohn’s disease) was asked about in separate questions; 793 women reported lupus, of whom 365 (0.57%) were counted as OA cases, and 1,780 women reported Crohn’s disease or ulcerative colitis, of whom only 825 (1.3%) were counted as OA cases, indicating that the general arthritis questions are good proxy indicators for OA
Covariates
A literature search was performed to acquire variables that have been strongly or moderately attributed to OA risk or protection. These included factors such as age, ethnicity, BMI, alcohol use, education, income, insurance status, smoking, postmenopausal hormone therapy, history of diabetes mellitus, and measures of physical activity. The WHI collected self-reported information or clinical measurements on most variables. Participant age was reported and categorized into three age ranges (50–59, 60–69, and 70–79). BMI (kg/m2) was calculated based on height and weight measurements at screening examinations. Women reported ethnicity in one of six groups: American Indian or Alaskan Native, Asian or Pacific Islander, African American, Hispanic or Latino, white (not Hispanic origin), or other. Women reporting other and those not specifying an ethnic group (n=1,813) were excluded from the analysis. Education status was categorized based on the highest grade finished in school. Annual household income was self-reported, and responses were categorized into six categories ranging from less than $20,000 to $100,000 or more. Insurance status was ascertained through questions regarding usual payment of medical care. Alcohol consumption categories included nondrinker, former drinker, and current drinker. Smoking status was classified into never smoker, past smoker, and current smoker. The women were asked about the number of days per week they participated in moderate (e.g., biking outdoors, using exercise machines) or strenuous (e.g., aerobics, swimming laps) exercise. Metabolic equivalent (MET) units were assigned, and continuous and categorical summary variables were created. Participants were asked about history of diabetes mellitus, current use of diabetic treatments, and postmenopausal hormone therapy (HT) use.
Statistical Analysis
Descriptive statistics were performed according to OA status, and chi-square was used to test statistical differences in the frequency of risk factors between the two groups. Continuous age, BMI, and the MET summary variable were also analyzed, but only categorical results are presented. Logistic regression was used to assess the association between the risk factors and the self-report of OA. Marginal analyses were performed using univariate logistic regression models, and the variables found to be significant (P<.20) were placed into the multivariate model. Backward elimination regression techniques were used to generate the final model, which included all variables significant at P<.05. Ethnic differences in the association between risk factors and OA were tested by examining the ethnic distribution of all variables as well as including an ethnicity and risk factor interaction term in logistic regression models. Ethnicity-specific odds ratios (ORs) were reported if significant interactions were found. All analyses were performed using Stata 10.0 (Statcorp, College Station, TX).
RESULTS
Demographics
At baseline, 63,699 (43.5%) women self-reported OA; 54,122 (83.9%) of those were non-Hispanic white, 5,955 (9.2%) African American, 2,117 (3.3%) Hispanic white, 1,203 (1.9%) Asian or Pacific Islander, and 302 (0.47%) American Indian or Alaskan Native. The OA group was significantly older, heavier, less educated, and less physically active and had a lower total family income than the reference group. Women with OA were significantly (P<.001) more likely to rate their overall health as fair or poor (11.5% and 1.1%, respectively) than the reference group (4.7% and 0.3%, respectively). Complete population characteristics can be found in Table 1.
Table 1.
Characteristics of Study Participants by Baseline Self-Reported Osteoarthritis
| Osteoarthritis | ||
|---|---|---|
| No | Yes | |
| N (%) | N (%) | |
| Total | 82,795 (56.5) | 63,699 (43.5) |
| Study Group* | ||
| CT | 36,747 (43.8) | 28,184 (41.5) |
| OS | 47,211 (56.2) | 39,725 (58.5) |
| Age (years) | ||
| 50–59 | 34,082 (40.6) | 16,784 (24.7) |
| 60–69 | 35,719 (42.5) | 32,280 (47.5) |
| 70–79 | 14,153 (16.9) | 18,845 (27.8) |
| Ethnicity | ||
| Non-Hispanic White | 69,256 (83.6) | 56,878 (84.9) |
| Hispanic White | 3,626 (4.4) | 2,321 (3.5) |
| African American | 6,862 (8.3) | 6,279 (9.4) |
| Asian | 2,734 (3.3) | 1,231 (1.8) |
| Native American | 317 (0.4) | 315 (0.5) |
| BMI (kg/m2) | ||
| <24.9 | 33,011 (39.7) | 20,338 (30.2) |
| 25.0–29.9 | 39,397 (35.2) | 23,047 (34.2) |
| 30.0–34.9 | 13,851 (16.7 | 13,845 (20.6) |
| 35.0–39.9 | 4,847 (5.8) | 6,391 (9.5) |
| ≥40 | 2,181 (2.6) | 3,698 (5.5) |
| Education | ||
| Less than high school | 3,560 (4.3) | 4,133 (6.1) |
| High School Diploma or GED | 13,486 (16.2) | 12,353 (18.3) |
| Some college/vocational/training school |
30,807 (37.0) | 26,187 (38.8) |
| College graduate or higher | 35,480 (42.6) | 24,737 (36.7) |
| Income | ||
| <$20,000 | 10,683 (13.6) | 12,454 (19.7) |
| $20,000–$34,999 | 17,602 (22.4) | 16,646 (26.3) |
| $35,000–$49,999 | 16,091 (20.5) | 13,039 (20.6) |
| $50,000–$74,999 | 16,902 (21.6) | 11,521 (18.2) |
| $75,000–$99,999 | 8,127 (10.4) | 4,862 (7.7) |
| $100,000+ | 9,026 (11.5) | 4,778 (7.5) |
| Insurance Status | ||
| Yes | 78,678 (94.6) | 64,856 (96.4) |
| No | 4,478 (5.4) | 2,388 (3.6) |
| Alcohol Consumption | ||
| Non Drinker | 8,855 (10.6) | 7,503 (11.1) |
| Past Drinker | 13,701 (16.4) | 14,009 (20.8) |
| Current Drinker | 60,862 (73.0) | 45,932 (68.1) |
| Smoking Status | ||
| Never Smoked | 42,878 (51.6) | 33,731 (50.3) |
| Past Smoker | 24,088 (41.0) | 29,017 (43.3) |
| Current Smoker | 6,101 (7.3) | 4,263 (6.4) |
|
Total Energy Expended in Physical Activity (METs*) |
||
| <1.25 | 13,993 (17.5) | 13,403 (20.8) |
| 1.25–5.49 | 15,733 (19.7) | 13,722 (21.3) |
| 5.50–11.66 | 16,144 (20.2) | 13,338 (20.7) |
| 11.67–20.9 | 16,012 (20.0) | 12,074 (18.8) |
| ≥21.0 | 18,134 (22.7) | 11,836 (18.4) |
| Diabetes Treatments (pills/shots) | ||
| No | 81,091 (96.7) | 60,962 (94.5) |
| Yes | 2,798 (3.3) | 3,520 (5.5) |
| Postmenopausal Hormone Therapy | ||
| Never Used | 37,697 (44.9) | 28,728 (42.3) |
| Past User | 12,303 (14.7) | 11,868 (17.4) |
| Current User | 33,890 (40.4) | 27,253 (40.2) |
METs = Metabolic Equivalent Unit
Associations of Arthritis Risk Factors
All variables tested in the marginal analysis were significant at the pre-set alpha level (P<.20) and were included in the full model. A significant linear trend with age was found, with odds of OA the highest in the group aged 70 to 79 (OR=2.69, 95% confidence interval (CI)=2.60–2.78), followed by those aged 60 to 69 (OR=1.81, 95% CI=1.76–1.86), with the group aged 50 to 59 serving as the reference (Table 2). Asian women had significantly lower odds of OA than non-Hispanic white women (OR=0.60, 95% CI=0.55–0.64). The marginal analysis revealed that African-American women had greater odds of OA (1.11, 95% CI=1.07–1.15) than non-Hispanic white women, although this relationship became statistically non-significant in the adjusted model. Native American women had a slightly greater odds of OA than non-Hispanic white women (OR=1.15, 95% CI=0.96–1.38).
Table 2.
Odds of Osteoarthritis for the Total Population and by Ethnicity
| White | Hispanic | African American |
Asian | American Indian |
Total Population* |
|
|---|---|---|---|---|---|---|
| OR (95%CI) |
OR (95%CI) |
OR (95%CI) |
OR (95%CI) |
OR (95%CI) |
OR (95%CI) |
|
| Age | ||||||
| 50–59 | Ref. | Ref. | Ref. | Ref. | Ref. | Ref. |
| 60–69 | 1.81 (1.76, 1.87) |
1.96 (1.72, 2.25) |
1.77 (1.62, 1.94) |
1.52 (1.26, 1.82) |
1.50 (0.98, 2.26) |
1.81 (1.76, 1.86) |
| 70–79 | 2.82 (2.27, 3.50) |
2.80 (2.25, 3.48) |
2.60 (2.28, 2.96) |
2.49 (2.00, 3.10) |
1.91 (1.10, 3.32) |
2.69 (2.60, 2.78) |
| BMI (kg/m2) | ||||||
| <24.9 | Ref. | Ref. | Ref. | Ref. | Ref. | Ref. |
| 25.0–29.9 | 1.24 (1.20, 1.28) |
1.18 (1.00, 1.39) |
1.23 (1.09, 1.39) |
1.27 (1.08, 1.49) |
1.21 (0.74, 1.99) |
1.23 (1.20, 1.27) |
| 30.0–34.9 | 1.55 (1.50, 1.61) |
1.47 (1.23, 1.75) |
1.69 (1.49, 1.92) |
1.16 (0.88, 1.55) |
1.59 (0.92, 2.73) |
1.55 (1.50, 1.60) |
| 35.0–39.9 | 2.12 (2.01, 2.24) |
1.83 (1.44, 2.33) |
2.26 (1.95, 2.62) |
1.66 (1.04, 2.67) |
2.71 (1.31, 5.58) |
2.11 (2.01, 2.22) |
| ≥40 | 2.71 (2.52, 2.92) |
2.47 (1.80, 3.41) |
3.31 (2.79, 3.91) |
3.22 (1.52, 6.84) |
4.22 (1.82, 9.77) |
2.80 (2.63, 2.99) |
| Education | ||||||
| Less than high school | Ref. | Ref. | Ref. | Ref. | Ref. | Ref. |
| High School Diploma or GED | 0.88 (0.81, 0.95) |
0.92 (0.75, 1.13) |
0.92 (0.78, 1.09) |
0.93 (0.65, 1.34) |
0.94 (0.48, 1.86) |
0.88 (0.82, 0.93) |
| Some college/training school | 0.88 (0.82, 0.94) |
0.98 (0.82, 1.17) |
0.83 (0.71, 0.96) |
0.86 (0.61, 1.21) |
0.91 (0.50, 1.63) |
0.87 (0.82, 0.92) |
| College graduate or higher | 0.84 (0.78, 0.91) |
0.85 (0.69, 1.04) |
0.83 (0.71, 0.98) |
0.89 (0.62, 1.26) |
0.62 (0.32, 1.19) |
0.84 (0.79, 0.89) |
| Income | ||||||
| <$20,000 | Ref. | Ref. | Ref. | Ref. | Ref. | Ref. |
| $20,000-$34,999 | 0.86 (0.83, 0.90) |
0.79 (0.67, 0.94) |
0.70 (0.63, 0.79) |
1.09 (0.84, 1.42) |
0.63 (0.38, 1.06) |
0.84 (0.81, 0.87) |
| $35,000-$49,999 | 0.81 (0.77, 0.84) |
0.76 (0.62, 0.92) |
0.67 (0.59, 0.76) |
1.10 (0.84, 1.44) |
0.56 (0.31, 1.01) |
0.79 (0.76, 0.82) |
| $50,000-$74,999 | 0.75 (0.71, 0.78) |
0.76 (0.61, 0.95) |
0.63 (0.55, 0.73) |
0.81 (0.62, 1.07) |
0.82 (0.44, 1.50) |
0.73 (0.70, 0.76) |
| $75,000-$99,999 | 0.72 (0.68, 0.76) |
0.55 (0.39, 0.76) |
0.49 (0.40, 0.59) |
0.96 (0.70, 1.31) |
0.52 (0.20, 1.36) |
0.69 (0.66, 0.73) |
| $100,000+ | 0.68 (0.64, 0.72) |
0.73 (0.51, 1.05) |
0.56 (0.44, 0.69) |
0.96 (0.69, 1.34) |
0.64 (0.22, 1.88) |
0.67 (0.63, 0.70) |
| Insurance Status | ||||||
| Yes | Ref. | Ref. | Ref. | Ref. | Ref. | Ref. |
| No | 0.62 (0.58, 0.67) |
0.58 (0.48, 0.69) |
0.73 (0.62, 0.85) |
0.83 (0.48, 1.42) |
1.58 (0.76, 3.29) |
0.64 (0.61, 0.69) |
| Alcohol Consumption | ||||||
| Non Drinker | Ref. | Ref. | Ref. | Ref. | Ref. | Ref. |
| Past Drinker | 1.17 (1.11, 1.23) |
0.96 (0.78, 1.17) |
1.24 (1.10, 1.41) |
1.04 (0.84, 1.27) |
0.87 (0.46, 1.65) |
1.16 (1.11, 1.21) |
| Current Drinker | 0.99 (0.94, 1.04) |
0.86 (0.72, 1.03) |
1.11 (0.98, 1.25) |
0.96 (0.81, 1.14) |
1.17 (0.64, 2.14) |
1.00 (0.96, 1.04) |
| Smoking Status | ||||||
| Never Smoked | Ref. | Ref. | Ref. | Ref. | Ref. | Ref. |
| Past Smoker | 1.08 (1.05, 1.11) |
1.13 (0.98, 1.30) |
1.17 (1.07, 1.28) |
1.02 (0.85, 1.21) |
1.12 (0.74, 1.68) |
1.09 (1.06, 1.11) |
| Current Smoker | 0.96 (0.91, 1.02) |
1.33 (1.05, 1.69) |
1.22 (1.07, 1.40) |
0.93 (0.62, 1.40) |
1.25 (0.64, 2.44) |
1.01 (0.96, 1.06) |
|
Total Energy Expended from Physical Activity (METS) |
||||||
| <1.25 | Ref. | Ref. | Ref. | Ref. | Ref. | Ref. |
| 1.25–5.49 | 0.95 (0.91, 0.99) |
0.93 (0.78, 1.11) |
1.03 (0.92, 1.15) |
0.96 (0.76, 1.22) |
0.61 (0.35, 1.07) |
0.96 (0.92, 0.99) |
| 5.50–11.66 | 0.95 (0.91, 0.99) |
0.87 (0.72, 1.05) |
0.95 (0.84, 1.07) |
0.90 (0.71, 1.14) |
0.87 (0.49, 1.56) |
0.94 (0.91, 0.98) |
| 11.67–20.9 | 0.91 (0.87, 0.95) |
0.85 (0.69, 1.04) |
0.86 (0.75, 0.98) |
0.98 (0.77, 1.24) |
0.69 (0.37, 1.30) |
0.91 (0.87, 0.94) |
| ≥21.0 | 0.82 (0.79, 0.86) |
0.74 (0.61, 0.91) |
0.71 (0.62, 0.81) |
0.90 (0.71, 1.14) |
0.68 (0.37, 1.22) |
0.81 (0.78, 0.85) |
| Diabetes Treatments (pills/shots) | ||||||
| No | Ref. | Ref. | Ref. | Ref. | Ref. | Ref. |
| Yes | 1.25 (1.16, 1.34) |
1.28 (1.00, 1.64) |
1.15 (1.01, 1.30) |
1.26 (0.93, 1.70) |
1.35 (0.74, 2.47) |
1.23 (1.16, 1.31) |
|
Postmenopausal Hormone Therapy |
||||||
| Never Used | Ref. | Ref. | Ref. | Ref. | Ref. | Ref. |
| Past User | 1.30 (1.25, 1.34) |
1.14 (0.94, 1.37) |
1.35 (1.20, 1.51) |
1.31 (1.04, 1.64) |
1.20 (0.70, 2.06) |
1.29 (1.25, 1.34) |
| Current User | 1.38 (1.35, 1.42) |
1.31 (1.14, 1.51) |
1.32 (1.20, 1.45) |
1.28 (1.09, 1.51) |
2.18 (1.41, 3.37) |
1.38 (1.34, 1.41) |
Adjusted for ethnicity
Women with a BMI of 40.0λkg/m2 or greater had a 2.80 (95% CI 2.63–2.99) greater odds of OA than women with a BMI less than 25.0λkg/m2. Higher levels of education and income were associated with lower odds of OA, and no clear association was seen with alcohol and smoking status. Women in the highest physical activity category had significantly lower odds of OA (OR=0.81, 95% CI=0.78–0.85) than women in the lowest category. The use of diabetic treatments was found to be associated with greater odds of OA (OR=1.23, 95% CI=1.16–1.31), and after adjusting for all variables, current HT usage was associated with greater odds of OA (OR=1.38, 95% CI=1.34–1.41). The adjusted ORs, including 95% CIs, for the total population can be found on Table 2.
Interaction Between Ethnicity and OA
Significant ethnic differences in the distribution of variables were seen, with African-American, American Indian, and Hispanic white women have more OA risk factors and less protective factors. For example, women from these ethnic groups were more likely to be obese than non-Hispanic white women (African American, 57.9%; American Indian, 51.0%; Hispanic, 41.9%; non-Hispanic white, 32.9%). These women also reported the least amount of physical activity (African American, 30.1%; American Indian, 29.5%; Hispanic, 27.9%; non-Hispanic white, 19.5%) and a higher percentage of women using diabetes treatments than non-Hispanic white (African American, 14.2%; American Indian, 15.9%; Hispanic, 8.6%; non-Hispanic white, 4.3%). Although the prevalence of OA increased with age, larger percentages of Hispanic, African American, and American Indian women reporting OA were in the group aged 50 to 59 than non-Hispanic white women (African American, 33.8%; American Indian, 36.4%; Hispanic, 39.3%; non-Hispanic white, 22.6%). The complete distribution of risk factors according to ethnicity can be found in Table 3.
Table 3.
Distribution of Risk Factors by Ethnicity in Women with OA
| White | Hispanic | African American |
Asian | American Indian |
|
|---|---|---|---|---|---|
| Total n (%) | 54,122 (85.0) | 2,117 (3.3) | 5,955 (9.4) | 1,203 (1.9) | 302 (0.5) |
| Study Group | |||||
| OS | 33,248 (61.4) | 1,207 (57.0) | 3,182 (53.4) | 785 (65.3) | 183 (60.6) |
| CT | 20,874 (38.6) | 910 (43.0) | 2,773 (46.6) | 418 (34.8) | 119 (39.4) |
| Age (years) | |||||
| 50–59 | 12,247 (22.6) | 831 (39.3) | 2,011 (33.8) | 310 (25.8) | 110 (36.4) |
| 60–69 | 26,041 (48.1) | 968 (45.7) | 2,774 (46.6) | 502 (41.7) | 131 (43.4) |
| 70–79 | 15,834 (29.3) | 318 (15.0) | 1,170 (19.6) | 391 (32.5) | 61 (20.2) |
| BMI (kg/m2) | |||||
| <24.9 | 17,373 (32.4) | 455 (21.8) | 761 (12.9) | 664 (55.5) | 62 (20.8) |
| 25.0–29.9 | 18,682 (34.7) | 760 (36.3) | 1,719 (29.1) | 391 (32.7) | 84 (28.2) |
| 30.0–34.9 | 10,519 (19.6) | 539 (25.8) | 1,629 (27.6) | 91 (7.6) | 78 (26.2) |
| 35.0–39.9 | 4,617 (8.6) | 213 (10.2) | 993 (16.8) | 32 (2.7) | 40 (13.4) |
| ≥40 | 2,513 (4.7) | 124 (5.9) | 797 (13.5) | 18 (1.5) | 34 (11.4) |
| Education | |||||
| Less than high school | 2,187 (4.1) | 562 (27.0) | 855 (14.5) | 89 (7.4) | 56 (18.7) |
| High School Diploma or GED | 10,035 (18.7) | 347 (16.7) | 906 (15.4) | 217 (18.1) | 51 (17.1) |
| Some college/vocational/training | 20,922 (38.9) | 757 (36.4) | 2,262 (38.5) | 413 (34.5) | 138 (46.2) |
| school | 20,622 (38.4) | 414 (19.9) | 1,857 (31.6) | 477 (39.9) | 54 (18.1) |
| College graduate or higher | |||||
| Income | |||||
| <$20,000 | 8,444 (16.7) | 775 (40.8) | 1,962 (36.1) | 157 (14.1) | 110 (38.7) |
| $20,000–$34,999 | 13,476 (26.6) | 461 (24.3) | 1,319 (24.3) | 252 (22.6) | 68 (23.9) |
| $35,000–$49,999 | 10,751 (21.2) | 301 (15.9) | 926 (17.0) | 235 (21.0) | 44 (15.5) |
| $50,000–$74,999 | 9,594 (19.0) | 230 (12.1) | 798 (14.7) | 227 (20.3) | 41 (14.4) |
| $75,000–$99,999 | 4,148 (8.2) | 72 (3.8) | 249 (4.6) | 129 (11.5) | 12 (4.2) |
| $100,000+ | 4,205 (8.3) | 60 (3.2) | 183 (3.4) | 117 (10.5) | 9 (3.2) |
| Insurance Status | |||||
| Yes | 52,403 (97.5) | 1,719 (84.3) | 5,347 (92.2) | 1,171 (98.2) | 265 (89.5) |
| No | 1,339 (2.5) | 319 (15.7) | 455 (7.8) | 21 (1.8) | 31 (10.5) |
| Alcohol Consumption | |||||
| Non Drinker | 5,000 (9.3) | 413 (19.8) | 995 (16.9) | 477 (39.8) | 45 (15.0) |
| Past Drinker | 10,098 (18.8) | 522 (25.0) | 2,163 (36.8) | 261 (21.8) | 86 (28.7) |
| Current Drinker | 38,743 (72.0) | 1,151 (55.2) | 2,714 (46.2) | 460 (38.4) | 169 (56.3) |
| Smoking Status | |||||
| Never Smoked | 26,516 (49.6) | 1,258 (60.5) | 2,792 (47.9) | 874 (73.0) | 139 (46.8) |
| Past Smoker | 23,587 (44.6) | 663 (31.9) | 2,389 (41.0) | 281 (23.5) | 127 (42.8) |
| Current Smoker | 3,101 (5.8) | 159 (7.6) | 643 (11.0) | 42 (3.5) | 31 (10.4) |
|
Energy Expended from Physical Activity (Mets) |
|||||
| <1.25 | 10,530 (19.5) | 584 (27.9) | 1,784 (30.1) | 228 (19.0) | 89 (29.5) |
| 1.25–5.49 | 11,250 (20.8) | 503 (24.0) | 1,497 (25.2) | 237 (19.7) | 57 (18.9) |
| 5.50–11.66 | 11,273 (20.9) | 406 (19.4) | 1,175 (19.8) | 242 (20.1) | 67 (22.2) |
| 11.67–20.9 | 10,587 (19.6) | 292 (13.9) | 776 (13.1) | 244 (20.3) | 40 (13.3) |
| ≥21.0 | 10,355 (19.2) | 312 (14.9) | 698 (11.8) | 252 (20.9) | 49 (16.2) |
| Diabetes Treatment (pills or shots) | |||||
| No | 51,773 (95.7) | 1,934 (91.4) | 5,103 (85.8) | 1,109 (92.3) | 248 (84.1) |
| Yes | 2,303 (4.3) | 182 (8.6) | 844 (14.2) | 93 (7.7) | 47 (15.9) |
| Postmenopausal Hormone Therapy | |||||
| Never Used | 21,691 (40.1) | 1,061 (50.2) | 3,514 (59.1) | 431 (35.9) | 138 (45.7) |
| Past User | 9,457 (17.5) | 299 (14.1) | 984 (16.6) | 201 (16.7) | 44 (14.6) |
| Current User | 22,932 (42.4) | 755 (35.7) | 1,446 (24.3) | 570 (47.4) | 120 (39.7) |
Several significant interaction terms were found, so the analysis was stratified according to ethnicity (Table 2). Slight differences in OA associations were seen according to ethnic group; the most noticeable was BMI. The odds of OA were much higher in American Indian (OR=4.22, 95% CI=1.82–9.77) and African American women (OR=3.31, 95% CI=2.79–3.91) in the highest BMI category than in non-Hispanic white women (OR=2.71, 95% CI=2.52–2.92). Current HT usage was significantly associated with greater odds of OA in all ethnic groups, although the stratified analysis revealed that the association was much higher in American Indian women (OR=2.18, 95% CI=1.47–3.47) than in the other ethnic groups, for example, Asian (OR=1.28, 95% CI=1.09–1.51) and white (OR=1.38, 95% CI=1.35–1.42).
DISCUSSION
OA is a highly prevalent condition in postmenopausal women, with 44% of the WHI participants reporting OA. The WHI self-reported prevalence of OA is similar to what was found in female participants of the Johnston County Osteoarthritis Project,[13] as well as to national prevalence estimates found previously.[3] The study confirmed several known OA risk factors and found some new and interesting results on the ethnic differences in OA risk factors.
The predominant risk factors confirmed in the analyses were age and BMI, with older age and higher BMI associated with greater odds of OA. Testing the interaction between BMI and ethnicity revealed a differential effect of obesity according to ethnicity on odds of OA. As shown in Table 2, American Indian and African-American women in the extreme obesity category had significantly greater odds of OA than non-Hispanic white women. Asian women had lower odds of OA in each BMI category than the other ethnic groups, although the odds of OA was dramatically greater in Asian women in the highest BMI category.
It is hypothesized that obesity plays a role in OA development and progression through two mechanisms; obesity increases dynamic stress on the joints, which leads to cartilage disruption, and obese people have a higher bone mineral density (BMD), which may increase subchondral bone stiffness and facilitate cartilage breakdown.[14] It has been shown that African Americans have higher BMD than other ethnic groups,[15] so higher BMD coupled with obesity may explain the greater prevalence of OA in the African-American population, as well as the poor joint health found in obese women from other ethnic groups.
Research has shown that African-American and Hispanic populations experience more-disabling effects of arthritis.[9] National population-based studies indicate substantially more disease activity and functional limitations in these groups than in non-Hispanic white Americans.[16] One study assessed ethnic differences in disease activity and found that African-American women had more pain and were considered more disabled than non-Hispanic white women.[17] This provides strong evidence that body weight and BMI may be a large contributing factor to the number and severity of OA symptoms, further elaborating the importance of postmenopausal women, especially African-American, Hispanic, and American Indian women, maintaining a healthy weight.
One unanticipated finding from this study was the greater odds of OA associated with HT use. OA has been linked to estrogen deficiency, and studies on postmenopausal hormone use and its relationship to OA have produced conflicting results. One study found that, in postmenopausal Italian women, users of estrogen replacement therapy had a 27% lower odds of physician-diagnosed OA than those who did not use estrogen replacements.[18] Several other studies have found HT use to be a protective factor,[19,20] although another study found that, after controlling for several risk factors (age, BMI, smoking, and exercise), women using postmenopausal estrogen had a five times greater risk of clinical hip OA, 30% higher knee OA risk, and 50% greater risk of hand OA.[4] Similarly, other studies have found HT to be a risk factor for OA.[21,22] The study found that past and current HT use was associated with 29% and 38% greater odds of OA. The methods used in the Italian study[18] are fairly similar to those used in the current study, although the characteristics of the Italian women were not comparable with those of the WHI population. Although results were similar, the women of one of the studies were all non-Hispanic white women from an upper class community and on average older than the women in the WHI.[4] The current study used self-reported OA and was not site specific. Further investigation of the quantity or duration of HT may provide clearer estimates of the effect of hormones on OA in this population, especially in the American Indian population.
American Indian women who reported current HT use at baseline had more than twice odds of OA than the population as a whole. A literature search was performed on American Indian women and postmenopausal hormone use, and only seven articles were found. The articles showed that HT use may contribute to greater risk of developing diabetes mellitus,[23] as well as higher levels of inflammatory factors,[24] which both have been shown to play a role in OA. The WHI Native American sample size is not as large as other groups (n=619), but several significant associations and interesting trends were found, signifying the need of further study of OA in this ethnic group.
Strengths and Limitations
The use of general arthritis data as a proxy for OA status is the most noticeable limitation of the study. The WHI arthritis question did not differentiate between arthritic conditions other than rheumatoid arthritis, and with more than 100 conditions considered in the broad category of arthritis, the associations found in this analysis could be weakened because the outcome may represent more than one condition. Nevertheless, OA remains the most common arthritic condition, especially in this age range. Conditions that are generally thought of as arthritic conditions other than OA are fairly rare in the general population. For example, rheumatoid arthritis has a prevalence of 1% in the general population, the prevalence of systemic lupus erythematosus (SLE) is estimated to be 40 to 50 cases per 100,000 persons, and the prevalence of spondylarthropathy (including ankylosing spondylitis, psoriatic arthritis, and inflammatory bowel disease) is estimated to be 2.1 cases per 1,000.[3] The WHI asked about rheumatic conditions such as SLE, Crohn’s disease, and ulcerative colitis in separate questions on the medical history questionnaire and specified OA in the medical history follow-up questionnaires, providing firm evidence that the initial arthritis questions were trying to assess the prevalence of OA and that the associations presented in this paper are indeed that of OA.
Using self-reported cases of the outcome is another limitation of this study. Although widely used, validation of this data collection method in OA has not been readily investigated. One study found that a rheumatologist could confirm 81% of self-reported OA cases.[25] The use of a cross-sectional study design limits the results of the analysis because a true temporal relationship between OA and the variables cannot be established. For example, does the lack of physical activity cause OA, or does the development of OA cause reduction in physical activity? The cross-sectional design of this study could also attribute to the discrepancy in HT results. The WHI is not representative of the entire U.S. population, and selection bias may cause under- or overestimation of the prevalence of OA as well as of the associations found. Although a major strength, the size of the WHI allows a statistical association to be found that may not necessarily be meaningful. As with any study, not controlling for all potential confounders and measurement error in data collection could bias the study results.
There are several noteworthy strengths of this study. This is the first study focusing on the prevalence of OA and its risk factors in a large multi-ethnic postmenopausal population. Approximately 20% of the WHI women are from ethnic minority backgrounds, and the women were recruited from 40 centers located across the United States. Because of the focus of the WHI, information on a variety of health information was collected, including almost all of the risk factors associated with OA. The large sample size provided sufficient power to observe important associations and trends in groups with smaller sample sizes. The large sample size also increased the ability to use multiple statistical methods to examine the relationship between the risk factors and OA. With the significant findings found in this study, further analyses using the high-quality data of the WHI is possible.
In conclusion, OA is a prevalent condition in postmenopausal women. This analysis revealed several differences in OA risk according to race or ethnicity in a group of highly motivated, healthy postmenopausal women. It is possible that there are greater ethnic differences in the general population, warranting further study of ethnic variation in frequencies and determining factors.
ACKNOWLEDGMENTS
Much gratitude is extended to the staff of the Health Aging Laboratory for all of their help. The WHI program is funded by the National Heart, Lung and Blood Institute, U.S. Department of Health and Human Services.
Funding Sources and related paper presentation:
The current paper is not associated with any funding agency and has not been previously presented. The WHI is funded by the National Heart Lung and Blood Institute.
Conflict of Interest: NCW has a research supplement grant from National Institute of Arthritis and Musculoskeletal and Skin Diseases (NIAMS). GKR is on contract to the Arizona Department of Health Services arthritis program. JL is on the Pfizer speakers bureau and has a research grant from Winston Pharmaceuticals. ZC has research grants from Eli Lilly and Company and the National Institutes of Health (NIAMS and National Institute on Aging).
Sponsor’s Role: The WHI was funded by the National Heart, Lung, and Blood Institute and provided minor edits on the manuscript before submission.
Appendix
SHORT LIST OF WHI INVESTIGATORS
Program Office
(National Heart, Lung, and Blood Institute, Bethesda, Maryland) Elizabeth Nabel, Jacques Rossouw, Shari Ludlam, Linda Pottern, Joan McGowan, Leslie Ford, and Nancy Geller.
Clinical Coordinating Center
(Fred Hutchinson Cancer Research Center, Seattle, WA) Ross Prentice, Garnet Anderson, Andrea LaCroix, Charles L. Kooperberg, Ruth E. Patterson, Anne McTiernan; (Wake Forest University School of Medicine, Winston-Salem, NC) Sally Shumaker; (Medical Research Labs, Highland Heights, KY) Evan Stein; (University of California at San Francisco, San Francisco, CA) Steven Cummings.
Clinical Centers
(Albert Einstein College of Medicine, Bronx, NY) Sylvia Wassertheil-Smoller; (Baylor College of Medicine, Houston, TX) Aleksandar Rajkovic; (Brigham and Women's Hospital, Harvard Medical School, Boston, MA) JoAnn Manson; (Brown University, Providence, RI) Annlouise R. Assaf; (Emory University, Atlanta, GA) Lawrence Phillips; (Fred Hutchinson Cancer Research Center, Seattle, WA) Shirley Beresford; (George Washington University Medical Center, Washington, DC) Judith Hsia; (Los Angeles Biomedical Research Institute at Harbor- UCLA Medical Center, Torrance, CA) Rowan Chlebowski; (Kaiser Permanente Center for Health Research, Portland, OR) Evelyn Whitlock; (Kaiser Permanente Division of Research, Oakland, CA) Bette Caan; (Medical College of Wisconsin, Milwaukee, WI) Jane Morley Kotchen; (MedStar Research Institute/Howard University, Washington, DC) Barbara V. Howard; (Northwestern University, Chicago/Evanston, IL) Linda Van Horn; (Rush Medical Center, Chicago, IL) Henry Black; (Stanford Prevention Research Center, Stanford, CA) Marcia L. Stefanick; (State University of New York at Stony Brook, Stony Brook, NY) Dorothy Lane; (The Ohio State University, Columbus, OH) Rebecca Jackson; (University of Alabama at Birmingham, Birmingham, AL) Cora E. Lewis; (University of Arizona, Tucson/Phoenix, AZ) Tamsen Bassford; (University at Buffalo, Buffalo, NY) Jean Wactawski-Wende; (University of California at Davis, Sacramento, CA) John Robbins; (University of California at Irvine, CA) F. Allan Hubbell; (University of California at Los Angeles, Los Angeles, CA) Lauren Nathan; (University of California at San Diego, LaJolla/Chula Vista, CA) Robert D. Langer; (University of Cincinnati, Cincinnati, OH) Margery Gass; (University of Florida, Gainesville/Jacksonville, FL) Marian Limacher; (University of Hawaii, Honolulu, HI) David Curb; (University of Iowa, Iowa City/Davenport, IA) Robert Wallace; (University of Massachusetts/Fallon Clinic, Worcester, MA) Judith Ockene; (University of Medicine and Dentistry of New Jersey, Newark, NJ) Norman Lasser; (University of Miami, Miami, FL) Mary Jo O’Sullivan; (University of Minnesota, Minneapolis, MN) Karen Margolis; (University of Nevada, Reno, NV) Robert Brunner; (University of North Carolina, Chapel Hill, NC) Gerardo Heiss; (University of Pittsburgh, Pittsburgh, PA) Lewis Kuller; (University of Tennessee, Memphis, TN) Karen C. Johnson; (University of Texas Health Science Center, San Antonio, TX) Robert Brzyski; (University of Wisconsin, Madison, WI) Gloria E. Sarto; (Wake Forest University School of Medicine, Winston-Salem, NC) Mara Vitolins; (Wayne State University School of Medicine/Hutzel Hospital, Detroit, MI) Susan Hendrix.
Footnotes
Author Contributions: NW played the major role in the study’s concept and design, analysis and interpretation of data, and preparation of the manuscript. GKR played a major role in the study’s concept, and manuscript preparation. JL played a role in the study’s concept, and manuscript preparation. ZC played a major role in the study’s concept, analysis and interpretation of data, and preparation of the manuscript.
References
- 1.National and state medical expenditures and lost earnings attributable to arthritis and other rheumatic conditions--United States, 2003. MMWR Morb Mortal Wkly Rep. 2007;56:4–7. [PubMed] [Google Scholar]
- 2.Nuki G. Osteoarthritis: a problem of joint failure. Z Rheumatol. 1999;58:142–147. doi: 10.1007/s003930050164. [DOI] [PubMed] [Google Scholar]
- 3.Lawrence RC, Helmick CG, Arnett FC, et al. Estimates of the prevalence of arthritis and selected musculoskeletal disorders in the United States. Arthritis Rheum. 1998;41:778–799. doi: 10.1002/1529-0131(199805)41:5<778::AID-ART4>3.0.CO;2-V. [DOI] [PubMed] [Google Scholar]
- 4.Von Muhlen D, Morton D, Von Muhlen CA, et al. Postmenopausal estrogen and increased risk of clinical osteoarthritis at the hip, hand, and knee in older women. J Womens Health Gend Based Med. 2002;11:511–518. doi: 10.1089/152460902760277868. [DOI] [PubMed] [Google Scholar]
- 5.Avci D, Bachmann GA. Osteoarthritis and osteoporosis in postmenopausal women: clinical similarities and differences. Menopause. 2004;11:615–621. doi: 10.1097/01.gme.0000119985.45613.7e. [DOI] [PubMed] [Google Scholar]
- 6.Dominick KL, Ahern FM, Gold CH, et al. Health-related quality of life among older adults with arthritis. Health Qual Life Outcomes. 2004;2:5. doi: 10.1186/1477-7525-2-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Seavey WG, Kurata JH, Cohen RD. Risk factors for incident self-reported arthritis in a 20 year followup of the Alameda County Study Cohort. J Rheumatol. 2003;30:2103–2111. [PubMed] [Google Scholar]
- 8.Oliver S, Hill J. Arthritis in the older person: Part 2. Nurs Older People. 2005;17(23):28–30. doi: 10.7748/nop2005.07.17.5.23.c2384. [DOI] [PubMed] [Google Scholar]
- 9.Jordan JM. Effect of race and ethnicity on outcomes in arthritis and rheumatic conditions. Curr Opin Rheumatol. 1999;11:98–103. doi: 10.1097/00002281-199903000-00003. [DOI] [PubMed] [Google Scholar]
- 10.Odutola J, Ward MM. Ethnic and socioeconomic disparities in health among patients with rheumatic disease. Curr Opin Rheumatol. 2005;17:147–152. doi: 10.1097/01.bor.0000151403.18651.de. [DOI] [PubMed] [Google Scholar]
- 11.Matthews KA, Shumaker SA, Bowen DJ, et al. Women's health initiative. Why now? What is it? What's new?[see comment] American Psychologist. 1997;52:101–116. doi: 10.1037//0003-066x.52.2.101. [DOI] [PubMed] [Google Scholar]
- 12.Hays J, Hunt JR, Hubbell FA, et al. The Women's Health Initiative recruitment methods and results. Annals of Epidemiology. 2003;13:S18–S77. doi: 10.1016/s1047-2797(03)00042-5. [DOI] [PubMed] [Google Scholar]
- 13.Jordan JM, Linder GF, Renner JB, et al. The impact of arthritis in rural populations. Arthritis Care Res. 1995;8:242–250. doi: 10.1002/art.1790080407. [DOI] [PubMed] [Google Scholar]
- 14.Cimmino MA, Parodi M. Risk factors for osteoarthritis. Semin Arthritis Rheum. 2005;34:29–34. doi: 10.1016/j.semarthrit.2004.03.009. [DOI] [PubMed] [Google Scholar]
- 15.Looker AC, Johnston CC, Jr, Wahner HW, et al. Prevalence of low femoral bone density in older U.S. women from NHANES III. J Bone Miner Res. 1995;10:796–802. doi: 10.1002/jbmr.5650100517. [DOI] [PubMed] [Google Scholar]
- 16.Shih VC, Song J, Chang RW, et al. Racial differences in activities of daily living limitation onset in older adults with arthritis: a national cohort study. Arch Phys Med Rehabil. 2005;86:1521–1526. doi: 10.1016/j.apmr.2005.02.009. [DOI] [PubMed] [Google Scholar]
- 17.Iren UT, Walker MS, Hochman E, et al. A pilot study to determine whether disability and disease activity are different in African-American and Caucasian patients with rheumatoid arthritis in St. Louis, Missouri, USA. J Rheumatol. 2005;32:602–608. [PubMed] [Google Scholar]
- 18.Parazzini F. Menopausal status, hormone replacement therapy use and risk of self-reported physician-diagnosed osteoarthritis in women attending menopause clinics in Italy. Maturitas. 2003;46:207–212. doi: 10.1016/s0378-5122(03)00193-2. [DOI] [PubMed] [Google Scholar]
- 19.Nevitt MC, Cummings SR, Lane NE, et al. Study of Osteoporotic Fractures Research Group. Association of estrogen replacement therapy with the risk of osteoarthritis of the hip in elderly white women. Arch Intern Med. 1996;156:2073–2080. [PubMed] [Google Scholar]
- 20.Spector TD, Nandra D, Hart DJ, et al. Is hormone replacement therapy protective for hand and knee osteoarthritis in women?: The Chingford Study. Ann Rheum Dis. 1997;56:432–434. doi: 10.1136/ard.56.7.432. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Sahyoun NR, Brett KM, Hochberg MC, et al. Estrogen replacement therapy and incidence of self-reported physician-diagnosed arthritis. Prev Med. 1999;28:458–464. doi: 10.1006/pmed.1998.0440. [DOI] [PubMed] [Google Scholar]
- 22.Sandmark H, Hogstedt C, Lewold S, et al. Osteoarthrosis of the knee in men and women in association with overweight, smoking, and hormone therapy. Ann Rheum Dis. 1999;58:151–155. doi: 10.1136/ard.58.3.151. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Zhang Y, Howard BV, Cowan LD, et al. The effect of estrogen use on levels of glucose and insulin and the risk of type 2 diabetes in american Indian postmenopausal women : the strong heart study. Diabetes Care. 2002;25:500–504. doi: 10.2337/diacare.25.3.500. [DOI] [PubMed] [Google Scholar]
- 24.Zhang Y, Howard BV, Cowan LD, et al. Associations of postmenopausal hormone therapy with markers of hemostasis and inflammation and lipid profiles in diabetic and nondiabetic american Indian women: the strong heart study. J Womens Health (Larchmt) 2004;13:155–163. doi: 10.1089/154099904322966137. [DOI] [PubMed] [Google Scholar]
- 25.March LM, Schwarz JM, Carfrae BH, et al. Clinical validation of self-reported osteoarthritis. Osteoarthritis Cartilage. 1998;6:87–93. doi: 10.1053/joca.1997.0098. [DOI] [PubMed] [Google Scholar]