Racial and ethnic differences in physical activity guidelines attainment among people at high risk of or having knee osteoarthritis


  • 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 Osteoarthritis Initiative.



This cross-sectional study examined racial/ethnic differences in meeting the 2008 United States Department of Health and Human Services Physical Activity Guidelines aerobic component (≥150 moderate-to-vigorous minutes/week in bouts of ≥10 minutes) among persons with or at risk of radiographic knee osteoarthritis (RKOA).


We evaluated African American versus white differences in guideline attainment using multiple logistic regression, adjusting for sociodemographic (age, sex, site, income, and education) and health factors (comorbidity, depressive symptoms, overweight/obesity, and knee pain). Our analyses included adults ages 49–84 years who participated in accelerometer monitoring at the Osteoarthritis Initiative 48-month visit (n = 1,142 with RKOA and n = 747 at risk of RKOA).


Two percent of African Americans and 13.0% of whites met the guidelines. For adults with and at risk of RKOA, significantly lower rates of guidelines attainment among African Americans compared to whites were partially attenuated by health factor differences, particularly overweight/obesity and knee pain (with RKOA: adjusted odds ratio [OR] 0.24, 95% confidence interval [95% CI] 0.08–0.72; at risk of RKOA: OR 0.28, 95% CI 0.07–1.05).


Despite known benefits from physical activity, attainment of the physical activity guidelines among persons with and at risk of RKOA was low. African Americans were 72–76% less likely than whites to meet the guidelines. Culturally relevant interventions and environmental strategies in the African American community targeting overweight/obesity and knee pain may reduce future racial/ethnic differences in physical activity and improve health outcomes.


Fueled by the aging population and the growing prevalence of osteoarthritis (OA), an epidemic of arthritis-associated disability is expected in the US over the next 2 decades (1). OA affects more than 27 million persons in the US (2). This number is expected to escalate due to the growing obesity epidemic (1, 3). Nearly two-thirds of obese adults are expected to develop knee OA during their lifetime (4). OA affecting the knee is a primary cause of disability (2, 5, 6).

Physical activity offers important benefits to adults, including persons with knee OA. Engaging in physical activity can reduce the risk of diabetes mellitus, cardiovascular disease, and stroke (7–9). Importantly among people with chronic disease, physical activity participation can prevent or delay disability (10). Randomized clinical trials show physiologic benefits from physical activity, including improved muscle strength, increased aerobic capacity, and flexibility (11–13). Recent federal United States Department of Health and Human Services (DHHS) Physical Activity Guidelines specifically include adults with arthritis to promote health benefits (14).

Although overall rates of disability among Americans have declined over time, racial/ethnic disparities in disability persist (15). Health promotion and equity are national priorities (16). The DHHS Physical Activity Guidelines serve as the benchmark for science-based recommendations on physical activity and health for Americans (14). Guideline attainment represents a potential step toward health equity for African Americans, particularly those having knee OA who are at high risk of developing disability. While earlier studies indicated that African Americans are less active than whites (17), activity levels in all racial ethnic groups are increasing (18); it is not known to what extent these changes translate into guideline attainment.

This study examines the extent of racial/ethnic differences between African Americans and whites in meeting the DHHS Physical Activity Guidelines aerobic recommendations. To guide public health and policy intervention, we further investigated modifiable health factors that may mediate racial/ethnic differences.

Significance & Innovations

  • Despite known benefits from physical activity, attainment of federally recommended physical activity levels among adults with and at risk of knee osteoarthritis (OA) was low.

  • Older African Americans were more than 70% less likely than their white counterparts with and at risk of knee OA to meet federally recommended physical activity levels, whose attainment can reduce poor health outcomes.

  • Racial/ethnic disparities in potentially life-threatening conditions may be reduced by decreasing barriers to being active, most notably obesity and knee pain, and through culturally targeted programs that promote physical activity.



The Osteoarthritis Initiative (OAI) is a longitudinal, prospective, observational study investigating the natural history of and risk factors for onset and progression of knee OA. The OAI recruited 4,796 adults, ages 45–79 years, who had or were at high risk of developing symptomatic radiographic knee OA (RKOA) at enrollment. All racial/ethnic groups were eligible to enroll. Annual OAI evaluations were done at 4 clinical sites: Baltimore, Maryland; Columbus, Ohio; Pittsburgh, Pennsylvania; and Pawtucket, Rhode Island, and are currently ongoing (online at http://www.oai.ucsf.edu/datarelease/About.asp). Institutional review board approval was obtained at the participating sites and at Northwestern University. Each participant provided written informed consent.

The OAI enrollment excluded individuals with rheumatoid or inflammatory arthritis; severe bilateral joint space narrowing in both knees or unilateral total knee replacement and contralateral severe joint space narrowing; history of or plans for bilateral knee replacement within 3 years; contraindications to undergo 3.0T magnetic resonance knee imaging; a positive pregnancy test; inability to provide a blood sample; use of aides other than a cane for more than 50% of ambulation; comorbid conditions that might interfere with study participation; or current participation in a double-blind randomized trial.

Annual knee radiographs used a fixed-flexion knee radiography protocol (19), including bilateral standing posteroanterior knee films with knees flexed to 20–30° and feet internally rotated 10°, using a Plexiglas positioning frame. RKOA was defined as Kellgren/Lawrence grade ≥2 at the 48-month visit (20).

Physical activity measurement.

A physical activity ancillary study collected accelerometer data on a subgroup of OAI participants with and at risk of RKOA at the OAI 48-month visit (21). Eligibility required a scheduled OAI 48-month visit between August 2008 and July 2010. Physical activity was monitored using ActiGraph GT1M uniaxial accelerometers. Trained research personnel gave uniform scripted in-person instructions to wear the accelerometer on a belt at the natural waistline on the right hip in line with the right axilla upon arising in the morning and continuously until retiring at night, except during water activities, for 7 consecutive days. Participants maintained a daily log to record time spent in water and cycling activities, which may not be fully captured by accelerometers. Participants returned the accelerometers to the research center, where data were downloaded using the manufacturer's software and checked for valid data recording.

Accelerometer output is an activity count, which is the weighted sum of the number of accelerations measured over a minute, where the weights are proportional to the magnitude of measured acceleration. Accelerometer data were analytically filtered using methodology validated in patients with rheumatic disease (21, 22). Nonwear periods were defined as ≥90 minutes with zero activity counts (allowing for 2 interrupted minutes with counts <100). A valid day of monitoring was defined as ≥10 wear hours in a 24-hour period. To provide reliable physical activity estimates, we restricted analyses to participants with 4–7 valid monitoring days (23). We calculated minutes of moderate-to-vigorous (MV; counts ≥2,020 minutes) physical activity occurring in bouts ≥10 minutes, with 1–2 minutes interruption below the threshold (23). Weekly totals were summed over 7 days or estimated as 7 times the average daily total for persons with 4–6 valid monitoring days. Using the 2008 DHHS Physical Activity Guidelines on aerobic activity, meeting the recommended level of physical activity was defined as ≥150 minutes/week bouted MV activity.


The following sociodemographic and health factors were considered as potential confounders based on their association with race/ethnicity and physical activity in previous studies (17, 24–28). Sociodemographic factors measured at baseline included race/ethnicity, age, sex, education years, and income. Race/ethnicity (African American, white, or other) was ascertained from self-report. Health factors were measured at the OAI 48-month visit. Body mass index (BMI), calculated from measured height (meters) and weight (kg) was classified into normal weight (BMI 18.5–24.9 kg/m2), overweight (BMI 25.0– 29.9 kg/m2), or obese (BMI ≥30 kg/m2). Presence of comorbidity was ascertained by Charlson Comorbidity Index >0 (29). High depressive symptoms were defined by a Center for Epidemiologic Studies Depression Scale (30) score of ≥16. Person-level knee pain (range 0–20, maximal pain) was evaluated from the maximum Western Ontario and McMaster Universities Osteoarthritis Index scores between the 2 knees. If a 48-month health factor was missing (1.6%, n = 31), the most recent annual assessment was used.

Study sample.

A total of 2,127 persons consented to accelerometer monitoring, representing 78.4% of the 2,712 eligible subcohort. By design, 2,084 OAI participants were not eligible due to clinic visits preceding the accelerometer study start date (1,543 cases) or were deceased/withdrew/did not return for the 48-month visit (541 cases). For analysis purposes, 52 participants reporting race/ethnicity other than African American or white and 186 participants having <4 valid days of accelerometer monitoring were excluded. The 1,889 participants comprising the analysis sample include 1,142 with RKOA in one or both knees and 747 at risk of RKOA (Figure 1).

Figure 1.

Flow chart of analytical sample of Osteoarthritis Initiative participants in an accelerometer study.

Statistical analysis.

OAI public data (31) were merged with accelerometer data. Analyses were performed separately for participants with and at risk of RKOA to provide results that can be generalized to these diagnostic groups. Compared with the analytical sample (n = 1,889), those not analyzed (n = 2,907) were disproportionally African Americans (20.2% versus 15.1%) and women (60.8% versus 55.0%). Recognizing such differences could influence our findings; weighted analyses based on Hogan recommendations were used to reflect the underlying OAI sample (32). All analytical results are weighted.

Descriptive statistics present age- and sex-adjusted percentages of meeting guidelines. Multiple logistic regression models of racial/ethnic differences adjusted for sociodemographic and health factors; an associated 95% confidence interval (95% CI) excluding 1 indicates statistical significance. Bivariate logistic regression models revealed no significant interactions between race/ethnicity and examined risk factors. Literature findings on racial/ethnic differences in physical activity guideline attainment were translated into odds ratios (ORs) to provide a consistent metric. Statistical testing used SAS software, version 9.2, and a nominal 5% alpha level.


The analysis sample of 1,889 participants (286 African Americans and 1,603 whites) included 1,142 with RKOA and 747 at risk of RKOA (Table 1). Among those participants with RKOA, African Americans, as compared to whites, were younger, more likely to be women, have lower income (<$50,000), have a lower education level (0–12 years), and have worse health, including higher rates of comorbidity, depressive symptoms, obesity, and greater knee pain severity. Similar patterns were observed for those at risk of RKOA.

Table 1. Characteristics of African Americans and whites with and at risk of RKOA participating in accelerometer monitoring*
CharacteristicsWith RKOA (n = 1,142)At risk of RKOA (n = 747)
African AmericanWhiteAfrican AmericanWhite
  • *

    Values are the percentage unless indicated otherwise. RKOA = radiographic knee osteoarthritis; BMI = body mass index.

  • Includes 6 participants with BMI 17.2–18.4 kg/m2.

  • Defined as a Center for Epidemiologic Studies Depression Scale score of ≥16.

  • §

    Person-level knee pain score was calculated using the maximum value of Western Ontario and McMaster Universities Osteoarthritis Index knee pain scores of the 2 knees; a higher number represents worse symptoms.

Participants, no.20693680667
Sociodemographic factors    
 Age, mean ± SD years62.9 ± 14.267.2 ± 13.861.7 ± 15.563.8 ± 15.0
 Years of education    
Health factors    
 Presence of comorbidity48.538.069.736.8
 BMI, mean ± SD kg/m232.0 ± 8.828.8 ± 7.428.8 ± 8.927.4 ± 7.1
 BMI normal (18.5–24.9 kg/m2)6.222.524.832.0
 BMI overweight (25.0–29.9 kg/m2)29.340.227.842.2
 BMI obese (≥30 kg/m2)64.537.447.425.8
 High depressive symptoms22.410.140.815.6
 Knee pain severity, mean ± SD§5.8 ± 8.43.0 ± 4.95.6 ± 8.12.2 ± 5.0

Factors associated with meeting DHHS Physical Activity Guidelines.

Overall, only 10.9% of OAI participants met the guidelines (with RKOA: 10.4%; at risk of RKOA: 11.8%) (Table 2). Racial/ethnic differences were substantial, with 2.0% of African Americans (with RKOA: 1.9%; at risk of RKOA: 2.3%) meeting guidelines compared to 13.0% of whites (with RKOA: 12.7%; at risk of RKOA: 13.4%) meeting guidelines. Guideline attainment rates, shown in Table 2, varied significantly by sociodemographic and health factors.

Table 2. Frequency of meeting 2008 DHHS Physical Activity Guidelines (aerobic) among adults with and at risk of RKOA*
CharacteristicsWith RKOAAt risk of RKOA
Sample size, no.Meeting guidelines, % (95% CI)Sample size, no.Meeting guidelines, % (95% CI)
  • *

    DHHS = US Department of Health and Human Services; RKOA = radiographic knee osteoarthritis; 95% CI = 95% confidence interval; BMI = body mass index.

  • Includes 6 participants with BMI 17.2–18.4 kg/m2.

  • Severe knee pain was classified as severe (score ≥10.9, based on the average score among adults with end-stage knee osteoarthritis awaiting total knee replacement) (48) versus not severe.

Overall1,14210.4 (8.6–12.2)74711.8 (9.5–14.1)
Sociodemographic factors    
  African American2061.9 (0.0–3.8)802.3 (0.6–8.3)
  White93612.7 (10.6–14.8)66713.4 (10.8–16.0)
 Age, years    
  49–6451814.0 (11.0–17.0)41112.2 (9.0–15.4)
  ≥656247.5 (5.4–9.6)33611.2 (7.8–14.6)
 Men51613.8 (10.8–16.8)33515.7 (11.8–19.6)
 Women6267.9 (5.8–10.0)4129.1 (6.3–11.9)
 Years of education    
  0–12 or missing1735.0 (1.8–8.2)778.8 (2.5–15.1)
  ≥1396911.6 (9.6–13.6)67012.2 (9.7–14.7)
  0–$49,9993755.2 (3.0–7.4)2267.1 (3.8–10.4)
  ≥$50,00071013.8 (11.3–16.3)50114.1 (11.1–17.1)
  Unknown579.7 (2.0–17.4)2023.2 (4.7–41.7)
Health factors    
 Presence of comorbidity    
  Yes4436.7 (4.4–9.0)2607.5 (4.3–10.7)
  No69912.8 (10.3–15.3)48714.9 (11.7–18.1)
 Depressive symptoms    
  High1264.9 (1.1–8.7)962.7 (0.9–7.4)
  Low1,01611.2 (9.3–13.1)65114.0 (11.3–16.7)
 BMI, kg/m2    
  Normal (18.5–24.9 kg/m2)22618.6 (13.5–23.7)25318.2 (13.4–23.0)
  Overweight (25.0–29.9 kg/m2)45513.6 (10.5–16.7)2889.7 (6.3–13.1)
  Obese (≥30 kg/m2)4613.9 (2.1–5.7)2067.9 (4.2–11.6)
 Severe knee pain    
  Yes581.8 (0.4–7.3)170.0
  No1,08410.9 (9.0–12.8)73012.5 (10.1–14.9)

African American versus white rates for meeting DHHS Physical Activity Guidelines.

The age- and sex-adjusted African American guideline attainment rate was less than one-seventh of that of whites (with RKOA: 2.2% versus 16.5%; without RKOA: 3.0% versus 16.6%).

Recognizing that differences in sociodemographic and health factors may contribute to racial/ethnic differences in guideline attainment, further analyses controlled for these factors (Table 3). After adjusting for demographics (age, sex, and OAI clinical site), African Americans, compared to whites, were 90% less likely to meet guidelines among adults with RKOA (OR 0.10, 95% CI 0.03–0.30) and were 81% less likely to meet guidelines among adults at risk of RKOA (OR 0.19, 95% CI 0.05–0.76). Further adjustment for education and income made little change. Estimates were further attenuated after adjustment for differences in health factors, including comorbidity, depressive symptoms, overweight/obesity, and knee pain (with RKOA: OR 0.24, 95% CI 0.08–0.72; at risk of RKOA: OR 0.28, 95% CI 0.07–1.05), but still remained statistically significant among adults with RKOA (African Americans were 76% less likely to meet guidelines).

Table 3. Racial/ethnic ORs (95% CIs) of physical activity guideline attainment (African Americans versus whites) among adults with and without RKOA*
Adjustment factorsWith RKOA (n = 1,142)Without RKOA (n = 747)
  • *

    ORs = odds ratios; 95% CIs = 95% confidence intervals; RKOA = radiographic knee osteoarthritis.

  • Health factors include comorbid conditions, depressive symptoms, body mass index, and pain severity.

 Demographics (age, sex, site)0.10 (0.03–0.30)0.19 (0.05–0.76)
 Demographics plus education/income0.13 (0.04–0.40)0.21 (0.05–0.87)
Sociodemographics plus all health factors0.24 (0.08–0.72)0.28 (0.07–1.05)

Multiple logistic regression models (not shown) among the RKOA group simultaneously adjusting for all risk factors showed, in addition to race/ethnicity, that factors significantly associated with lower odds of meeting guidelines included older age (OR 0.96, 95% CI 0.94–0.98), female sex (OR 0.57, 95% CI 0.37–0.88), being obese (OR 0.20, 95% CI 0.10–0.40) or overweight (OR 0.60, 95% CI 0.36–1.00), and having greater knee pain (OR 0.86, 95% CI 0.79–0.95). Among persons at risk of RKOA, significant factors included older age (OR 0.96, 95% CI 0.94–0.99), female sex (OR 0.49, 95% CI 0.30–0.80), high depressive symptoms (OR 0.28, 95% CI 0.10–0.80), being obese (OR 0.37, 95% CI 0.20–0.69) or overweight (OR 0.48, 95% CI 0.27–0.86), and having greater knee pain (OR 0.85, 95% CI 0.76–0.96).

To understand which potentially modifiable health factors most strongly explain the observed racial/ethnic differences, we separately adjusted the models for each health factor. For this purpose, the reference is the sociodemographics-adjusted racial/ethnic OR in Table 3 (with RKOA: OR 0.13; at risk of RKOA: OR 0.21). For the RKOA group, there was little change in racial/ethnic differences related to comorbidities or depressive symptoms. However, greater knee pain and higher BMI each attenuated the racial/ethnic differences (from the reference OR 0.13 to OR 0.16, 95% CI 0.05–0.50 and to OR 0.18, 95% CI 0.06–0.56, respectively). Among adults at risk of RKOA, the only individual health factor that attenuated racial/ethnic differences was greater knee pain (from the reference OR 0.21 to OR 0.29, 95% CI 0.07–1.13).


This study investigated racial/ethnic differences in meeting the aerobic component of the DHHS Physical Activity Guidelines among adults with and at risk of RKOA. Our results showed that despite the known benefits from physical activity, only 1 of 50 African Americans and less than 1 of 8 whites met guidelines when objectively assessed via accelerometer. After adjusting for sociodemographic and health factors, African American rates for meeting guidelines were less than one-third of their white counterparts; these differences remained significant in the RKOA subgroup and substantial but insignificant in adults at risk of RKOA.

The current 2008 DHHS Physical Activity Guidelines are less strict than earlier recommendations, allowing the activity to be acquired over a week without daily specifications (≥150 minutes of MV activity/week) and taking into consideration persons who do a mix of both moderate and vigorous physical activity. When the guidelines were applied to the general adult population surveyed by the 2007 Behavioral Risk Factor Surveillance System (BRFSS) (33) and the 2005–2006 National Health and Nutrition Examination Survey (NHANES) (25) self-reported physical activity data, similar racial differences were found (OR range 0.54–0.63). In contrast, accelerometer-measured objective physical activity data of the NHANES sample did not show strong racial/ethnic differences in meeting the guidelines (OR 0.93) (25). However, other NHANES findings indicate that racial/ethnic differences were age specific (23). African Americans age <60 years compared to whites spent more time, on average, in objectively measured moderate and vigorous physical activity, but among adults age >60 years this trend was reversed. Those NHANES results corroborate the current results documenting racial/ethnic differences in an older chronic-disease cohort. These studies in the general population showed strong racial/ethnic disparities in meeting guidelines when based on self-reported physical activity data, but showed ambiguous results when physical activity was objectively measured.

The literature on racial/ethnic differences in meeting physical activity guidelines among adults with arthritis is limited. Self-reported physical activity data from the 2000 (34), 2001 (35), and 2003 BRFSS (36) were applied to the 2003 physical activity guidelines (≥30 minutes of moderate-intensity physical activity on ≥5 days per week or ≥20 minutes of vigorous-intensity physical activity on ≥3 days per week) (37) in arthritis populations. All 3 studies showed African Americans were less likely than whites to meet the 2003 guidelines (OR range 0.49–0.74). To our knowledge, no studies from arthritis populations used objective physical activity measurements to assess racial/ethnic differences in guideline attainment. Our study fills this gap by demonstrating large racial/ethnic differences in meeting the current guidelines using accelerometer-monitored physical activity in adults with RKOA.

To guide public health and policy intervention, we examined modifiable health factors that may mediate racial/ethnic differences. Obese/overweight status and greater knee pain partially attenuated racial/ethnic differences for adults with RKOA. Overweight/obese and pain are frequently reported as factors associated with inactivity (26, 27) and as barriers to participating in MV physical activity (25). In fact, pain is the most common barrier to physical activity among adults with arthritis (28). Our sample was consistent with the literature that finds that African Americans with OA are heavier than whites and report substantially higher levels of knee pain (24). These findings are relevant to potential public health action targeting overweight/obesity and knee pain in the African American community to reduce future racial/ethnic differences in guideline attainment.

Importantly, racial/ethnic differences in meeting guidelines remained substantial and significant in adults with RKOA after controlling for all assessed sociodemographic and health factors. These persistent differences support the need for future exploration of racial/ethnic barriers to activity. In recent systematic reviews, unsafe neighborhoods, lack of facilities, lack of child care, and inflexible work environments were the most commonly reported barriers to physical activity among African Americans (38, 39). Common facilitators of physical activity among African Americans included social support, availability of structured/group exercise programs, positive health benefits, a sense of well-being, and weight loss (38). Siddiqi et al (38) concluded that “African American adults clearly stress the need for targeted programs (e.g., through faith-based interventions) and the availability of safe and accessible facilities and places that are conducive to physical activity.” The Centers for Disease Control and Prevention (CDC) recommend 6 structured, group exercise programs as safe and effective (reduce pain, increase function, improve mood and quality of life, and delay disability) for adults with arthritis (40). These programs have been delivered in churches, local community centers, etc., in both urban and rural environments and work sites.

Churches can significantly contribute to the health and well-being of African Americans because of their role in communication, social and spiritual support, and the ability to provide safe local facilities (41). Among a sample of African American churches in South Carolina, 42% offered physical activity programs. Church members who reported having physical activity programs at their church were significantly more likely to meet physical activity recommendations (42). Offering evidence-based, low-cost structured programs in churches may overcome barriers to physical activity reported by African Americans. In addition, a tested CDC community-wide health communication campaign, Physical Activity: The Arthritis Pain Reliever, targeted to African Americans and whites that promotes physical activity for managing arthritis symptoms, increased knowledge of the benefits of physical activity for arthritis (43). Health communication campaigns paired with availability of structured community-based physical activity programs may positively increase physical activity participation in African American communities.

Lower levels of physical activity among African Americans compared with whites, particularly among adults with OA who are at high risk for multiple poor health outcomes, is a significant public health issue. African Americans are more likely than whites to develop obesity- and physical inactivity–related chronic conditions such as type 2 diabetes mellitus, stroke, cardiovascular disease, and hypertension (44–47). Not only does physical activity improve arthritis symptoms, it lowers the risk of developing these serious conditions (7, 8, 37). Racial/ethnic disparities in potentially life-threatening conditions may be reduced through culturally targeted programs that promote physical activity and reduce barriers to activity, most notably obesity and knee pain.

This study had substantial strengths, which include the large sample size, objective accelerometer physical activity assessment, radiographic verification of knee OA, the application of current DHHS Physical Activity Guidelines, and the age and sex diversity of this cohort. Accelerometers capture all activity, including occupational, household, and transportation, which may not be well-captured using self-report instruments. However, there are limitations worth noting. Accelerometers lack information on context of physical activity (e.g., transportation, leisure), which may inform culturally relevant activities for interventions. The accelerometer used cannot capture water activities and may underestimate activities with minimal vertical acceleration/deceleration, such as cycling. However, diary information indicated that the median time this sample spent in water and cycling activities was 0 minutes/day (interquartile range 0.0–3.4 minutes/day), so the underestimate is negligible. In addition, the OAI is not a population-representative sample. Therefore, our results here cannot be generalized to the US population.

Despite benefits from physical activity, attainment of the 2008 DHHS Physical Activity Guidelines was low for all groups. African Americans were even less likely than whites to meet the guidelines; this relationship held among persons with or at risk of RKOA. After controlling for differences in sociodemographics and health factors, substantial racial/ethnic differences remained. These disparities were partially mediated by differences in knee pain severity and obesity status. Focused efforts in African Americans to address knee pain severity and weight status may improve physical activity participation and lead to better health outcomes.


All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be submitted for publication. Dr. Song 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 conception and design. Song, Hochberg, Chang, Hootman, Manheim, Lee, Dunlop.

Acquisition of data. Song, Hochberg, Dunlop.

Analysis and interpretation of data. Song, Hochberg, Chang, Hootman, Manheim, Lee, Semanik, Sharma, Dunlop.


The Osteoarthritis Initiative private funding partners (Merck Research Laboratories, Novartis Pharmaceuticals Corporation, GlaxoSmithKline, and Pfizer) had no role in the study design, data collection, data analysis, or writing of this manuscript. Publication of this article was not contingent on the approval of these sponsors.


We thank the OAI participants, the OAI study sites and their diligent research staff, and recognize the valuable project support of Ms Leilani Lacson and Ms Emily Arntson.