The Association of Diabetes With Knee Pain Locations, Pain While Walking, and Walking Speed: Data From the Osteoarthritis Initiative

Abstract

Objective

Osteoarthritis (OA) and diabetes mellitus (DM) often coexist and can result in negative outcomes. DM can affect pain and walking speed in people with knee OA; however, the impact of DM on OA is understudied. The purpose of this study was to investigate the association between diabetes and knee pain locations, pain severity while walking, and walking speed in people with knee OA.

Methods

A cross-sectional analysis was used. Data from 1790 individuals from the Osteoarthritis Initiative (mean [SD] age = 69 [8.7] years) with knee pain were included and grouped into knee OA and diabetes (n = 236) or knee OA only (n = 1554). Knee pain locations were categorized as no pain, localized pain, regional pain, or diffuse pain. Knee pain during a 20-m walk test was categorized as no pain, mild, moderate, or severe knee pain. Walking speed was measured using the 20-m walk test. Multinomial and linear regression analyses were performed.

Results

Diabetes was associated with regional knee pain (odds ratio [OR] = 1.77; 95% CI = 1.01–3.11). Diabetes was associated only with moderate (OR = 1.78; 95% CI = 1.02–3.10) or severe (OR = 2.52; 95% CI = 1.01–6.28) pain while walking. Diabetes was associated with decreased walking speed (B = −0.064; 95% CI = −0.09 to −0.03).

Conclusions

Diabetes was associated with regional knee pain but not with localized or diffuse knee pain and was associated with moderate to severe knee pain while walking and slower walking speed in people with knee OA.

Impact

Clinicians can use a knee pain map for examining knee pain locations for people with diabetes and knee OA. Knee pain during walking and walking speed should be screened for people with knee OA and diabetes because of the influence of diabetes on these parameters in this population.

Lay Summary

Diabetes might be associated with specific knee pain locations, pain during activities such as walking, and reduced walking speed in people with knee OA.

Osteoarthritis (OA) and diabetes mellitus (DM) are common chronic diseases affecting approximately 14% and 10% of the general population, respectively.^1–4 OA is characterized by joint pain, which is one of the leading causes of disability and is the main reason for seeking medical intervention.⁴ Many factors and comorbidities, including age and DM, affect pain and are associated with increased knee pain severity.^5,6

DM is characterized by high blood glucose due to disturbances in insulin metabolism leading to hyperglycemia, which often leads to systemic changes in body organs including joints.⁷ Another consequence of hyperglycemia is the production of advanced glycation end products that can accumulate in any part of the body, including joints, and can increase cartilage stiffness and bone fragility.⁸ DM and OA share some risk factors that can explain their higher prevalence,⁹ but their pathophysiological relationship is still unclear. A number of studies have shown a significant association between DM and OA incidence and progression.^9–13 Further, DM has been shown to be an independent risk factor for progression of hip or knee OA and is associated with negative outcomes following joint replacement.^10,14–18 Previous evidence suggests that systemic metabolic syndromes such as obesity and DM can play a role in the pathophysiology of OA at any joint.¹⁹

Pain is a complex phenomenon that involves multiple domains and factors, including joint pathology, psychological distress, and neurophysiology.²⁰ The perception of pain can be the result of structural abnormalities, joint or systemic inflammation, and/or malalignment. In many cases, systemic inflammation can be attributed to chronic conditions such as DM, and hyperglycemia is associated with proinflammatory and prodegradative effects on OA. Under hyperglycemia, advanced glycation end products can accumulate in cells and joints, modifying joint properties and causing stiffness, resistance, and cartilage degradation.^7,8,21 These physiological changes due to DM and/or hyperglycemia could explain the higher pain severity experienced by people with OA and DM when compared with people with OA only.^5,17,22,23

Pain is often considered a fifth vital sign that can be characterized by severity and location. Previous evidence has suggested that patients with knee OA in more than one location often express pain that is more diffuse or spreads to larger areas,²⁴ supporting the hypothesis of central sensitization.²⁵ Diffuse knee pain has been associated with more physical dysfunction, higher levels of anxiety, more widespread pain, neuropathic-like pain, and more pain when compared with localized pain in people with knee OA.^24,26–29 Therefore, management of knee OA depends on pain location. For example, previous guidelines recommended progressive hip strengthening and home-based exercise for specific localized medial knee OA pain.^30,31 On the other hand, patients with diffuse pain or central sensitization require an interdisciplinary approach that incorporates interventions such as cognitive behavioral therapy,³² mind–body exercise,^33,34 and use of specific footwear. An understanding of the association between DM and OA is necessary to identify specific pain patterns to establish the appropriate intervention. In addition, managing DM with good glycemic control can influence pain in people with OA, and should be incorporated into the interventional approach.²²

Only a few studies have reported higher pain severity in adults with DM and OA.^5,17,23,35 However, these studies examined subgroups of individuals with OA such as those with end-stage knee OA before arthroplasty⁵ or an erosive form of hand OA²³ that is considered severe.³⁶ There is currently a lack of evidence regarding pain during daily activities such as walking, and the possible negative impact of that pain on walking speed and functionality.

Walking speed is often considered a sixth vital sign³⁷ and an important predictor of disability³⁸ and mortality³⁹ in older adults. Previous research has shown that a decline in walking speed is associated with poor health outcomes,⁴⁰ and associated separately with knee OA or DM. People with hip or knee OA walk slower than others with healthy joints.^41,42 Furthermore, adults with DM demonstrate slower walking speeds when compared with healthy individuals.^43,44 However, although knee OA or DM are associated with slower walking speed, the impact of DM on walking speed in people with knee OA remains unknown.

Understanding the impact of DM on pain experience and walking speed in people with knee OA is valuable because it will help establish interdisciplinary approaches for this population. Therefore, the aims of this study were: (1) to examine the association between DM and knee pain locations, and (2) to investigate the impact of DM on knee pain while walking and on walking speed in individuals with knee OA and knee pain. We hypothesized that DM will be associated with diffuse knee pain, higher pain severity during the 20-m walk test, and decreased walking speed in people with knee OA.

Methods

Study Design

This study was a cross-sectional analysis of data from the Osteoarthritis Initiative (OAI) at 96 months follow-up. The OAI is an ongoing, multisite longitudinal study in the United States that includes 4796 participants with or at risk of knee OA, with the goal of investigating the impact of knee OA over time to better understand prevention and treatment strategies. This study was approved by the University of California, San Francisco (UCSF) Institutional Review Board and its affiliates (approval number: FWA00000068). Institutional Review Board approval was also obtained from all 4 clinical sites located at Brown University in Providence, Rhode Island; Ohio State University in Columbus, Ohio; University of Maryland/Johns Hopkins University joint center in Baltimore, Maryland; and the University of Pittsburgh in Pennsylvania. All participants provided informed consent before screening and/or recruitment, and each participant signed a consent form before enrollment. For this study, we used OAI data for patients with knee OA (https://data-archive.nimh.nih.gov/oai/).

Participants

The OAI protocol includes diverse groups of individuals aged 45 to 79 years. Participants are divided into 3 cohorts; a progression cohort (n = 1389 participants) who have symptomatic knee OA with presence of both osteophytes and frequent knee symptoms in at least 1 knee; an incidence cohort (n = 3285 participants) with nonsymptomatic knee OA but an increased risk of symptomatic OA in at least 1 knee; and a control cohort (n = 122 participants) with no symptomatic or radiographic knee OA and no risk of OA. Risk factors for the incidence cohort include having knee symptoms, being overweight, having a knee injury, knee surgery, family history, Heberden node, repetitive knee bending, or aged between 70 and 79 years. Participants in the control cohort have no pain, no aching or stiffness in either knee, no radiographic findings, and none of the risk factors mentioned above. In this study, we used data from participants with knee OA at baseline measured by radiographic composite OA grade (≥2) and reporting knee pain in at least 1 knee over 12 months according to the following question: “During the past 12 month, have you had this pain, aching, or stiffness in your right/left knee?” Previous longitudinal studies have used these questions.^45,46 The Figure shows the flow of included participants (n = 1790). We excluded participants with missing DM status and further divided into knee OA and DM (n = 236) or knee OA only without DM (n = 1554) groups.

Figure.

Flow chart of participant selection. OA = osteoarthritis; OAI = Osteoarthritis Initiative.

Measures

Self-reported DM from the Charlson Comorbidity Index was used to categorize participants with and without DM. Simply, participants were asked whether they had been diagnosed with DM and provided a yes/no response. Previous research has reported the validity and reliability of self-reported DM using the Charlson Comorbidity Index.^47,48

A knee pain map was administered at 24 and 96 months follow-up; we selected data from the 96-month follow-up due to a smaller amount of missing data for DM status and knee pain location (7.6%). The knee pain map is an interviewer-administered survey that identifies painful areas of the knee. All interviewers were trained to use the knee pain map with high reliability (K = 0.7–1.0).⁴⁹ Areas were defined as localized when participants indicated an area that hurt using 1 or 2 fingers, regional when participants indicated an area that hurt using their hand over the region, or diffuse when participants said that it hurts everywhere. This procedure was performed with the participant seated at the edge of an examination table with their legs bent over the edge. Trained interviewers identified and recorded locations using a drawing of a knee divided into specific locations used in a previous study.²⁴ This study reported that, “pain was recorded as being in one of seven local areas (superior medial, medial joint line, inferior medial, patella, superior lateral, lateral joint line, inferior lateral), one of four regional areas (medial, patella, lateral, back) or as diffuse pain that cannot be localized or regionalized. If participants reported more than four local areas of pain or more than two regions of pain in a knee, their pain in that knee was classified as ‘diffuse.’ Participants were also allowed to identify one location and one non-overlapping region of pain.”²⁴ These locations were further categorized into 4 categories, including no pain, localized pain, regional pain, or diffuse pain. If the participant reported knee pain locations in both knees, the most symptomatic knee was included for analysis.

Knee pain while walking was measured in each knee immediately after the completion of a 20-m walk test via the following question: “Please rate the maximum amount of pain you experienced while walking, from 0 (no pain) to 10 (severe pain).” The most symptomatic knee was selected for this analysis. Knee pain while walking was further categorized as: no pain, mild pain (1–3), moderate pain (4–6), or severe pain (7–10).⁵⁰

Usual pace walking speed was measured during a 20-m walk test using the average of 2 trials. Certified site assessors used a standardized protocol to measure walking speed for the OAI participants. Walking speed was computed by dividing the distance (20 m) by the time (seconds) needed to complete the test. Participants were instructed to wear their usual footwear and used assistive devices if needed.

Age, gender, and body mass index (BMI) were included as covariates because previous research has shown their association with knee pain severity and location.^24,29 Participants were classified as having depression symptoms if they scored 16 or more on the Center for Epidemiologic Studies depression scale, and this was included as a covariate.⁵¹ Radiographic evidence at baseline for tibiofemoral knee OA using the OAI composite OA grade was also included as a covariate.

Statistical Analyses

Data and descriptive statistics are presented as means for continuous variables and percentages for categorical variables. All analyses were performed using SPSS for Macintosh, version 25.0 (SPSS Inc, Chicago, IL, USA). Significance level was set at .05.

Two multinomial logistic regression analyses were used. One examined the impact of DM on knee pain locations and the other examined the impact of DM on knee pain during the 20-m walk test. Knee pain locations were categorized as no pain, localized pain, regional pain, or diffuse pain, whereas knee pain during the 20-m walk test was categorized as no pain, mild pain, moderate pain, or severe pain. Two models were created, with DM as the predictor and knee pain while walking and knee pain locations as dependent variables. Model 1 was adjusted for age, gender, and radiographic OAI composite OA grade, whereas model 2 was adjusted for these variables with the addition of BMI and depression symptoms. These covariates have been associated with knee pain and/or locations in people with knee OA.^52,53 Odds ratios (ORs) with associated CIs were calculated for each model and for each category of the outcome variable. The reference categories for both outcome variables were set as no pain location and no pain while walking, respectively.

Multiple linear regression analyses were used to examine the impact of DM on walking speed during the 20-m walk test. Two models were created with DM as the predictor and walking speed (meters per second) as the dependent variable. Model 1 was adjusted for age, gender, and radiographic OAI composite OA grade, whereas model 2 was adjusted for these variables in addition to pain while walking, BMI, and depression symptoms. A listwise deletion was used for some models due to missing data for some variables (eg, BMI). Linear regression results were presented as unstandardized beta coefficients (B) and 95% confidence interval (CI).

Role of the Funding Source

The funders played no role in the design, conduct, or reporting of this study.

Results

As illustrated in the Figure, this study included a total of 1790 participants. Of these 236 (13.18%) self-reported DM. Of the participants with DM, 60.2% were female, 20.5% had depression symptoms, and the mean BMI was 32.3 kg/m². Table 1 shows participants’ characteristics for both the sample and DM subsample.

Table 1.

Participants’ Demographics and Clinical Characteristics^a

Characteristic	Total Sample (n = 1790)	Diabetes Subsample (n = 236)
Mean age, y (SD)	69.65 (8.77)	69.93 (8.43)
Female, n (%)	967 (59.3)	142 (60.2)
BMI,b kg/m2	29.81 (5.23)	32.30 (5.09)
Depression symptoms,b yes, n (%)	215 (13.6)	46 (20.5)
Knee pain location, n (%)	1333 (74.5)	197 (83.5)
No pain	207 (15.5)	23 (11.7)
Localized	432 (32.4)	68 (34.5)
Regional	366 (27.5)	61 (31.0)
Diffuse	328 (24.6)	45 (22.8)
Mean walking speed,b m/s (SD)	1.24 (0.22)	1.12 (0.21)
Knee pain while walking,b n (%)	1350 (75.4)	191 (80.9)
No pain	902 (66.8)	111 (58.1)
Mild pain	331 (24.5)	48 (25.1)
Moderate pain	89 (6.6)	23 (12.0)
Severe pain	28 (2.1)	9 (4.7)

^a BMI = body mass index. ^bStatistically significant at .05 level using χ² or independent t test.

Results from the multinomial logistic regression analyses are presented in Table 2. The final adjusted model (model 2, n = 1300) revealed that DM remained significantly associated with regional knee pain (OR = 1.77; 95% CI = 1.01–3.11) when compared with no DM and no pain, and after controlling for age, gender, BMI, depression symptoms, and OA grade. Diffuse and localized knee pain were not significantly associated with DM in either model.

Table 2.

Multinomial Logistic Regression for the Association Between Diabetes and Knee Pain Locations

Knee Pain Locations	Model 1a (n = 1317)	P	Model 2b (n = 1300)	P
Localized	1.64 (.97–2.80)c	.06	1.65 (.95–2.87)c	.07
Regional	1.76 (1.03–3.01)c	.04	1.77 (1.01–3.11)c	.04
Diffuse	1.37 (.78–2.40)c	.27	1.15 (.64–2.07)c	.65

^a Model 1 adjusted for age, sex, and knee composite grade.

^b Model 2 adjusted for age, sex, knee composite grade, body mass index, and depression symptoms. Reference category was set as no pain.

^c 95% CI.

Results for knee pain while walking using the multinomial logistic regression analyses are presented in Table 3. The final adjusted model (model 2, n = 1316) showed that DM remained significantly associated with moderate (OR = 1.78; 95% CI = 1.02–3.10) and severe (OR = 2.52; 95% CI = 1.01–6.28) pain while walking when compared with no DM and no pain while walking, and after controlling for age, gender, BMI, depression symptoms, and OA grade.

Table 3.

Multinomial Regression for the Association Between Diabetes and the Maximum Amount of Pain Experienced in Worst Knee During the 20-m Walk

Knee Pain While Walking	Model 1a (n = 1331)	P	Model 2b (n = 1316)	P
Mild knee pain	1.26 (.87–1.84)c	.22	1.08 (.73–1.60)c	.70
Moderate knee pain	2.52 (1.49–4.24)c	.001	1.78 (1.02–3.10)c	.04
Severe knee pain	3.50 (1.45–8.44)c	.005	2.52 (1.01–6.28)c	.04

^a Model 1 adjusted for age, gender, and knee composite grade.

^b Model 2 adjusted for age, gender, knee composite grade, body mass index, and depression symptoms. Reference category was set as no pain.

^c 95% CI.

The results of the linear regression analysis to examine the impact of DM on walking speed are presented in Table 4. The final adjusted model (model 2, n = 1316) showed that DM was significantly associated with decreased walking speed (B = −0.064; 95% CI = −0.09 to −0.03) after controlling for age, gender, knee pain while walking, BMI, depression symptoms, and OA grade.

Table 4.

Linear Regression Analyses for Walking Speed Measured by 20-m Walk Test

Diabetes vs No Diabetes	Model 1a (n = 1331)			Model 2b (n = 1316)
Walking speed	B	SE	P	B	SE	P
	−0.12	0.016	<.001	−0.064	0.015	<.001
R 2	0.17			0.30

^a Model 1 adjusted for age, gender, and knee composite grade.

^b Model 2 adjusted for age, gender, knee composite grade, body mass index, and depression symptoms. Reference category was set as no pain.

Discussion

This study examined the impact of DM on knee pain locations, knee pain while walking, and walking speed in individuals with knee OA. We found that DM was specifically associated with regional knee pain, with moderate and severe knee pain while walking, and with decreased walking speed.

The most frequent knee pain location reported in this study was localized knee pain, followed by regional and diffuse knee pain. These findings were consistent with a previous study,²⁴ but contradictory to others.^28,54 The consistency of our findings with those of Thompson et al²⁴ could be attributable to the fact that we used the same OAI database sample, albeit at different time points (ie, 24 months follow-up vs 96 months follow-up). In contrast, however, other studies have found that diffuse knee pain was the most frequent pain location.^28,54 This discrepancy could be due to the use of different definitions of knee pain location or different knee pain mapping positions (eg, sitting vs standing). Further, patients referred to larger hospitals could have a greater degree of OA severity and pain that might result in more central, diffuse knee pain.⁵⁴

Contrary to our hypothesis, DM was associated with regional knee pain but not diffuse knee pain after controlling for covariates such as BMI and depression. This lack of association between DM and diffuse knee pain requires further research, because the identification of specific knee pain locations associated with DM could bolster understanding of how comorbidities influence pain in people with knee OA. This could allow researchers and clinicians to clarify possible mechanisms linking DM and OA, and to quantify progression or changes in pain location. Recognizing potential associations between DM and regional knee pain could facilitate identification of clinical phenotypes of OA, help monitor knee pain, and improve the implementation of effective treatment approaches incorporating DM management. Diffuse knee pain might be associated with chronic comorbidities such as metabolic syndrome and/or diabetes,²⁴ but our study did not find such an association. In this study, adults with DM were about 1.77 times more likely to report regional pain when compared with those without diabetes and no knee pain. Only a few studies have examined risk factors associated with knee pain locations using a knee pain map.^24,26 Although Thompson et al²⁴ reported an association between higher BMI and all knee pain locations, the association was greater for diffuse knee pain. However, this study did not examine the risk of DM, and the only link between this study and our study is the metabolic syndrome for obesity not DM. Use of pain medication is another potential confounding factor that was not measured in the current and previous studies. Regional knee pain might be the result of pain triggers from multiple sources, including richly innervated tissues such as the periosteum, synovium, muscle, joint capsule, and ligament.⁵⁵ These structures can be affected by DM and chronic hyperglycemia. DM is also associated with low-grade systemic inflammation leading to joint inflammation and pain,⁵⁶ and evidence has linked pain to increased levels of inflammatory markers in people with DM and OA when compared with those with knee OA only.^5,57 Zhang et al⁵⁷ showed that DM was associated with an increased production of advanced glycation end products, suggesting another pathway by which DM can affect joint pain and inflammation. In addition, Eitner et al⁵ have indicated that specific inflammatory biomarkers such as interleukin-6, are associated with more pain, and this association was dependent on DM status.

Moderate and severe pain while walking was associated with DM in our study, with adults with DM being about 1.77 to 2.5 times more likely to have moderate or severe pain when compared with individuals without DM and without knee pain while walking. It is known that knee OA is associated with joint stiffness and altered joint biomechanics that can affect pain.^58,59 Recent evidence has indicated that medial knee loading is associated with more pain while walking among people with knee OA. DM has been also been linked to altered gait parameters and biomechanics.^60–66 Therefore, in addition to the low-grade systemic inflammation associated with DM, its effects on joint biomechanics^67–69 could also influence pain while walking. Our findings were consistent with previous research that found higher pain severity in people with DM and knee OA compared with knee OA only.⁵ However, unlike this study, we measured pain severity while walking.

We found that walking speed was significantly slower in people with DM, independent of knee pain while walking, with a mean walking speed of 1.12 m/s in people with DM and knee pain compared with 1.24 m/s in the full sample. This exceeds the 0.1 m/s threshold for a clinically meaningful difference in walking speed.⁷⁰ We also found that DM was associated with decreased walking speed after controlling for other covariates (B = −0.064; 95% CI = −0.09 to 0.03). These findings were consistent with a previous report in adults with DM.⁴⁴ Volpato et al⁴⁴ found that individuals with DM had decreased walking speed on a 400-m walk test when compared with those without DM (B = −0.053 m/s). Another study by Kera et al⁷¹ reported similar findings, although the mean walking speed was higher than our study for people with DM (1.32 m/s). Previous research on walking speed and knee OA has not controlled for knee pain while performing a walking speed test. However, our study was able to control for knee pain while walking to examine the influence of DM on walking speed independent of pain. Recent research on walking speed in individuals with knee OA showed that Japanese women had a lower walking speed than men.⁷² In our study, age and gender were included as covariates in the analyses, and the association between DM and walking speed remained significant. Glycemic control can also influence OA severity, and our recent work found that poorer glycemic control (measured by increased hemoglobin A_1c) was associated with increased pain severity in people with localized OA after controlling for medication usage and other confounders.²² Thus, because pain perception appears to be altered based on DM status and glycemic control, physical functions such as walking speed could serve as a better treatment guide than pain severity.

This study has potential limitations that should be considered. First, self-reported DM is a key variable in this study, and it is possible that the presence of undiagnosed DM, prediabetes, denial, or lack of awareness resulted in inaccurate or underestimated reporting. Also, this self-report did not distinguish between type 1 and type 2 DM. However, approximately 95% of diabetes cases are classified as type 2. Second, the cross-sectional design of this study limits interpretation of the results, and causal relationships between DM and knee pain location, pain while walking, or walking speed cannot be drawn. Longitudinal studies are needed to explore the decline in gait speed in those with DM and OA over time. Third, this study used a wide range of categories for variables such as BMI that might limit the generalizability of the results. Fourth, neuropathic status and the length of time since DM diagnosis were not captured in this database. Because DM duration and/or neuropathy can affect musculoskeletal integrity and sensation, disruptions in afferent perception and joint load attenuation could influence our results. Thus, the associations between the measures in this study should be interpreted cautiously. Our data are limited by the fact that our main outcome (knee pain location) categorizes any of many areas into a single location (eg, localized). However, this approach is necessary for the analysis of such an outcome. Finally, we were unable to account for potential confounders such as pain medication and hemoglobin A_1c in this study. Future research should investigate whether good versus poor glycemic control influences the associations between DM and knee pain and walking speed. Our recent work in a small sample has highlighted a possible relationship between an increase in hemoglobin A_1c value and pain severity in patients with OA and DM.²² It is crucial that physical functions in people with OA and DM are examined, because previous research has shown an association between DM and OA progression.^10,17 However, further work is needed to determine how glycemic control might mediate relationships between DM and physical function.

In conclusion, we found that DM was associated with regional knee pain but not diffuse or localized knee pain in people with knee OA who had experienced pain during the past 12 months. DM was also associated with moderate to severe knee pain while walking and with slower walking speed in people with knee OA. DM is known to cause musculoskeletal and somatosensory damage that can affect pain location and walking performance. Identification of the impact of DM on knee pain location, pain while walking, and walking speed could help clinicians develop more appropriate interventions. Given that slower gait speed is associated with an elevated risk of falls, we suggest that clinicians include walking speed assessments for patients with DM and knee OA to help mitigate this risk.

Author Contributions

Concept/idea/research design: A.M. Alenazi, J. Rucker, N.K. Sharma, S.M. Bindawas, P.M. Kluding

Writing: A.M. Alenazi, M.M. Alshehri, S. Alothman, B.A. Alqahtani, J. Rucker, N.K. Sharma, S.M. Bindawas, P.M. Kluding

Data analysis: A.M. Alenazi, S. Alothman

Project management: A.M. Alenazi

Consultation (including review of manuscript before submitting): S. Alothman, B.A. Alqahtani, J. Rucker, S.M. Bindawas, P.M. Kluding

Ethics Approval

This study was approved by the University of California, San Francisco (UCSF) Institutional Review Board (IRB) and its affiliates (approval no. FWA00000068). IRB approval was also obtained from all 4 clinical sites located at Brown University in Providence, Rhode Island; Ohio State University in Columbus, Ohio; University of Maryland/Johns Hopkins University joint center in Baltimore, Maryland; and the University of Pittsburgh in Pittsburgh, Pennsylvania. All participants provided informed consent before screening and/or recruitment, and each participant signed a consent form before enrollment.

Funding

This work was supported by the Deanship of Scientific Research at Prince Sattam bin Abdulaziz University, Alkharj, Saudi Arabia. The Osteoarthritis Initiative (OAI) is a public-private partnership comprised of 5 contracts (N01-AR-2-2258, N01-AR-2-2259, N01-AR-2-2260, N01-AR-2-2261, N01-AR-2-2262) funded by the US National Institutes of Health (NIH), a branch of the Department of Health and Human Services, and conducted by the OAI Study Investigators. Private funding partners include Merck Research Laboratories, Novartis Pharmaceuticals Corporation, GlaxoSmithKline, and Pfizer Inc. Private sector funding for the OAI is managed by the Foundation for the National Institutes of Health.

Disclosures and Presentations

The authors completed the ICMJE Form for Disclosure of Potential Conflicts of Interest and reported no conflicts of interest.

This manuscript was prepared using an OAI public use data set and does not necessarily reflect the opinions or views of the OAI investigators, NIH, or the private funding partners.

Parts of this work were presented as a poster at the 2018 American Congress of Rehabilitation Medicine, Dallas, Texas, October 1-3, 2018, and at the 2019 American College of Sports Medicine Annual Meeting, Orlando, Florida, May 28-June 1, 2019.