Abstract
Osteoarthritis (OA) is a prevalent musculoskeletal disorder worldwide, and its onset is closely associated with obesity. The prevalence of obesity among American adults continues to rise; however, the impact of different types of obesity (general obesity vs abdominal obesity) on OA remains to be thoroughly investigated. This study utilized data from the National Health and Nutrition Examination Survey in the United States between 2005 and 2018, incorporating 21,709 participants aged 20 years and above. The association between OA and both body mass index (BMI) and waist circumference (WC) was analyzed using multivariate logistic regression. The models were adjusted for demographic characteristics, lifestyle factors, and metabolic factors, and restricted cubic spline analysis was employed to examine the dose–response relationship. The fully adjusted multivariable analysis demonstrated significant positive associations of both BMI (odds ratio [OR] = 1.06, 95% confidence interval [CI]: 1.05–1.07) and WC (OR = 1.02, 95% CI: 1.02–1.03) with OA prevalence. Participants with obesity (BMI ≥ 30 kg/m²) exhibited an adjusted OR of 2.72 (95% CI: 1.67–4.43) for OA versus the underweight group (BMI < 18.5 kg/m²), and 2.29 versus the normal-weight group (BMI 18.5–24.9 kg/m²). Those with abdominal obesity showed an adjusted OR of 1.57 (95% CI: 1.35–1.82) compared to the normal WC group. Subgroup analyses revealed significant interaction effects for race (interaction P = .002) and smoking status (interaction P < .001), with no significant interactions observed for other demographic or metabolic indicators (all P > .05). Both general obesity and abdominal obesity are independently positively associated with the prevalence of OA, and the strength of these associations increases with higher levels of BMI and WC. Public health interventions should focus particularly on obesity control, especially the management of visceral fat accumulation, to reduce the burden of OA.
Keywords: abdominal obesity, BMI, multivariate logic analysis, NHANES, obesity, osteoarthritis, WC
1. Introduction
Osteoarthritis (OA) is the most common musculoskeletal disorder worldwide, characterized clinically by joint pain, swelling, stiffness, and functional impairment. These symptoms significantly affect patients’ quality of life and mobility.[1] In the United States, the prevalence of OA continues to rise and has become one of the primary causes of chronic pain and functional impairment among middle-aged and elderly populations.[2] Globally, approximately 9.6% of men and 18% of women aged 60 and over are affected by OA. The condition often progresses gradually and is difficult to reverse. In severe cases, it can lead to physical disability and loss of function.[3] The pathogenesis of OA is complex and involves multiple factors, including age, gender, obesity, and joint biomechanical abnormalities. Among these, obesity serves both as a systemic metabolic risk factor and as a promoter of local degeneration through increased joint loading.[4] As a growing public health concern globally, obesity is on the rise among American adults. According to data from the Centers for Disease Control and Prevention, approximately 42.4% of American adults are considered obese, with significant variations in prevalence across different racial and gender groups.[5]
Body mass index (BMI) and waist circumference (WC) are important indicators for assessing obesity. However, BMI primarily reflects the overall degree of obesity and does not accurately distinguish between fat and muscle tissue, nor does it reflect the distribution of body fat.[6,7] WC, as an indicator of abdominal obesity, can more directly reflect the accumulation of visceral fat.[8] Research indicates that abdominal fat, particularly visceral fat, participates in cartilage degeneration and synovial inflammation by secreting inflammatory mediators, closely linking it to the systemic inflammatory mechanisms of OA.[9] Although previous studies based on National Health and Nutrition Examination Survey (NHANES) data have confirmed the overall association between BMI and OA, evidence suggests that abdominal obesity significantly increases the risk of OA even among individuals with normal BMI.[10,11] This study, for the 1st time, reveals the population heterogeneity of obesity-related OA by comprehensively assessing 2 obesity indicators (BMI and WC) and their interaction effects with race and smoking status (interaction P < .05).
This study hypothesizes that both general obesity and abdominal obesity are independently associated with the risk of OA, and analyzing the relationship between obesity indicators (BMI and WC) and OA risk may help establish a more precise OA risk stratification system and inform individualized management strategies for populations with different obesity phenotypes. Based on this, this study conducted a cross-sectional analysis using NHANES data from 2005 to 2018 to clarify the relationship between obesity/abdominal obesity and OA.
2. Materials and methods
2.1. Data sources and study population
The NHANES is a 2-year, nationally representative cross-sectional survey project implemented by the National Center for Health Statistics, which is under the Centers for Disease Control and Prevention. NHANES employs a multistage, stratified, probability sampling design and includes multidimensional health data such as demographic information, questionnaire surveys, physical examinations, and laboratory tests to systematically assess the health and nutritional status of the American public. It explores the relationships between diseases, lifestyle factors, and socioeconomic status, providing scientific evidence for the formulation of national public health policies. Data can be accessed from the NHANES website (https://www.cdc.gov/nchs/nhanes/index.htm). All surveys are conducted with informed consent from participants and strictly adhere to ethical standards and data privacy protection regulations.
This study integrates and analyzes datasets from 7 consecutive cycles of NHANES between 2005 and 2018, involving a total of 70,190 individuals. The exclusion criteria were as follows: individuals under 20 years old (n = 30,441); individuals lacking data on arthritis, BMI, or WC (n = 4016); pregnant individuals (n = 547); individuals with rheumatoid arthritis or other types of arthritis (n = 5906); individuals missing relevant covariates (n = 7571). Ultimately, 21,709 eligible participants were included in the final analysis (Fig. 1).
Figure 1.
A flow chart of participants screening in NHANES 2005–2018. BMI = body mass index, NHANES = National Health and Nutrition Examination Survey.
2.2. Exposure variable
BMI and WC are commonly used indicators in clinical and epidemiological studies for assessing obesity. They play complementary roles in determining obesity types and predicting health risks. BMI, calculated as an individual’s weight in kilograms divided by the square of their height in meters, serves as a quantitative measure of overall obesity. According to the international standards established by the World Health Organization, BMI is categorized into 4 clinical classifications: underweight (<18.5 kg/m²), normal weight (18.5–24.9 kg/m²), overweight (25–29.9 kg/m²), and obesity (≥30 kg/m²).[12] This classification system effectively assesses the relationship between total body fat and the risk of chronic diseases. WC, as a key indicator of abdominal obesity, directly reflects the degree of abdominal and visceral fat accumulation. The clinical cutoff values for WC differ by gender: men with a WC ≥ 102 cm and women with a WC ≥ 88 cm are considered to have abdominal obesity. A substantial amount of evidence-based medical research indicates that when WC exceeds these thresholds, the risk of developing metabolic abnormalities, cardiovascular diseases, and diabetes complications significantly increases for individuals.[13–15] The combined use of these 2 indicators, BMI and WC, allows for a comprehensive assessment of obesity levels and distinguishes between different types of fat distribution. This provides more precise reference criteria for clinical interventions and health management.
2.3. Outcome variable
The assessment of arthritis was conducted using a self-reported questionnaire (MCQ160A), which asked, “Has a doctor or other health professional ever told you that you have arthritis?” If the response was “yes,” the participant was then asked to answer a follow-up question: “What type of arthritis is it?” Based on the responses to these 2 questions, participants were categorized into either the OA group or the no arthritis group.
Although previous studies have shown that conclusions obtained using self-reporting methods are consistent with clinical diagnoses up to 81% of the time[16] this approach may miss asymptomatic or undiagnosed OA patients. This nondifferential misclassification could lead to an underestimation of effect sizes, meaning the observed strength of the association between BMI/WC and OA might be slightly lower than the actual level. While this does not affect the judgment of statistical significance, appropriate caution should be exercised when interpreting the strength of the association. Future studies need to further enhance diagnostic accuracy through objective measures such as imaging examinations.
2.4. Covariates
The questionnaire, in addition to surveying OA, BMI, and WC, also examined various confounding factors, including gender, age, race, education level, poverty income ratio (PIR), smoking status, alcohol consumption, diabetes, hypertension, and hyperlipidemia. For sociodemographic covariates, race was categorized into Mexican American, other Hispanic, non-Hispanic White, non-Hispanic Black, and other race. Education levels were classified as follows: <9th grade, 9–11th grade, high school grade/general educational development or equivalent, some college or associate degree, and college graduate or above. Smoking status was defined as “no” if the participant had smoked fewer than 100 cigarettes in their lifetime, and as “yes” if they had smoked 100 or more cigarettes in their lifetime.[17] Participants’ alcohol consumption was assessed based on two 24-hour dietary intake recall data. If alcohol consumption was reported in either of the recalls, the participant was classified as an alcohol user.[18] In health-related covariates, diabetes was defined based on fasting glucose measurements, physician diagnosis, the use of glucose-lowering medications, and insulin usage. Participants were classified as having diabetes if they met any one of the following criteria: being informed by a doctor that they have diabetes, having a fasting blood glucose level of ≥126 mg/dL (7.0 mmol/L), having a glycated hemoglobin (HbA1c) level of ≥6.5%, or currently taking glucose-lowering medications or using insulin.[19] Hypertension was defined as an average systolic blood pressure of ≥140 mm Hg or a diastolic blood pressure (DBP) of ≥90 mm Hg, regardless of whether there is a physician diagnosis or record of medication treatment. In addition, if both the average systolic blood pressure and DBP are below these thresholds but the respondent reported having been diagnosed with hypertension by a doctor or is currently taking prescribed antihypertensive medications, they were also classified as hypertensive.[20] Hyperlipidemia was defined as having any of the following criteria: triglycerides ≥ 150 mg/dL, total cholesterol ≥ 200 mg/dL, low-density lipoprotein ≥ 130 mg/dL, or high-density lipoprotein ≤ 40 mg/dL for men and ≤50 mg/dL for women. Participants who reported using cholesterol-lowering medications were also classified as having hyperlipidemia.[21]
2.5. Statistical analysis
Data processing and analysis were performed using R 4.4.2, along with Zstats v0.90 (www.medsta.cn/software). A P < .05 (2-sided) was considered statistically significant. To ensure that the estimates are representative of the U.S. population, all analyses were weighted according to the complex sampling design of NHANES. Continuous variables were presented as mean ± SD (weighted), and categorical variables were expressed as weighted percentages. First, a descriptive statistical analysis was conducted to summarize the baseline characteristics of the participants. To assess the association between obesity/abdominal obesity and OA, multivariable logistic regression models were used to calculate odds ratios (ORs) and 95% confidence intervals (CIs). Three covariate-adjusted models were evaluated: model 1 was unadjusted; model 2 was adjusted for gender, race, and age; model 3 was adjusted for gender, race, age, education, alcohol, hypertension, smoke, hyperlipidemia, diabetes, marital, and PIR. A trend test was performed to examine the dose–response relationship between clinically stratified BMI and WC categories and the risk of OA. Subsequently, restricted cubic spline (RCS) regression was employed to evaluate the potential nonlinear associations between BMI/WC and OA risk, along with threshold effect analysis. Finally, subgroup analyses were conducted stratifying by gender, age, race, education level, marital status, PIR, smoking status, alcohol consumption, diabetes, hypertension, and hyperlipidemia. Interaction effects were also tested across these subgroups.
3. Results
3.1. Baseline characteristics of participants
Table 1 presents the baseline characteristics of all individuals with and without OA. A total of 21,709 participants were included in the final analysis. The mean age of the study participants was 45.61 ± 0.26 years, with an average BMI of 28.63 ± 0.08 kg/m² and a mean WC of 98.27 ± 0.22 cm. Among individuals with OA, 63.74% were female, 62.5% were married, and >32% had attained a college education or higher. Except for DBP, all other study variables showed statistically significant differences between groups (P < .05). Furthermore, the comparison of baseline characteristics between participants included in the analysis and those excluded due to missing covariates is presented in Supplementary Table S1, Supplemental Digital Content, https://links.lww.com/MD/P956.
Table 1.
Clinical characteristics of the study participants
| Variable | Total (n = 21,709) | non-OA (n = 19,029) | OA (n = 2680) | P value |
|---|---|---|---|---|
| Gender, n (%) | <.001 | |||
| Male | 11,104 (50.26) | 10,095 (52.41) | 1009 (36.26) | |
| Female | 10,605 (49.74) | 8934 (47.59) | 1671 (63.74) | |
| Age, mean (SE) | 45.61 ± 0.26 | 43.21 ± 0.24 | 61.26 ± 0.30 | <.001 |
| Marital, n (%) | <.001 | |||
| Married | 11,426 (56.45) | 9918 (55.52) | 1508 (62.50) | |
| Widowed | 1253 (4.17) | 841 (3.01) | 412 (11.77) | |
| Divorced | 2203 (9.76) | 1815 (9.25) | 388 (13.05) | |
| Separated | 679 (2.20) | 606 (2.27) | 73 (1.77) | |
| Never married | 4218 (18.73) | 4010 (20.48) | 208 (7.36) | |
| Living with partner | 1930 (8.68) | 1839 (9.47) | 91 (3.54) | |
| Race, n (%) | <.001 | |||
| Mexican American | 3439 (8.48) | 3232 (9.37) | 207 (2.69) | |
| Other Hispanic | 2045 (5.28) | 1867 (5.74) | 178 (2.27) | |
| Non-Hispanic White | 9566 (69.19) | 7842 (66.86) | 1724 (84.36) | |
| Non-Hispanic Black | 4198 (9.75) | 3800 (10.34) | 398 (5.93) | |
| Other race | 2461 (7.29) | 2288 (7.68) | 173 (4.75) | |
| Education, n (%) | .016 | |||
| <9th Grade | 1872 (4.30) | 1682 (4.48) | 190 (3.11) | |
| 9–11th Grade | 2830 (9.52) | 2515 (9.65) | 315 (8.61) | |
| High school grad/GED or equivalent | 4854 (22.29) | 4250 (22.38) | 604 (21.70) | |
| Some college or AA degree | 6597 (31.66) | 5718 (31.26) | 879 (34.27) | |
| College graduate or above | 5556 (32.23) | 4864 (32.22) | 692 (32.32) | |
| PIR, mean (SE) | 3.10 ± 0.03 | 3.07 ± 0.03 | 3.25 ± 0.06 | .002 |
| Alcohol, n (%) | .002 | |||
| Yes | 6753 (35.72) | 6014 (36.30) | 739 (31.97) | |
| No | 14,956 (64.28) | 13,015 (63.70) | 1941 (68.03) | |
| Smoke, n (%) | <.001 | |||
| Yes | 9417 (43.68) | 7978 (42.30) | 1439 (52.65) | |
| No | 12,292 (56.32) | 11,051 (57.70) | 1241 (47.35) | |
| SBP, mean (SE) | 120.93 ± 0.19 | 120.00 ± 0.19 | 126.97 ± 0.49 | <.001 |
| DBP, mean (SE) | 70.72 ± 0.20 | 70.80 ± 0.21 | 70.19 ± 0.36 | .085 |
| Hypertension, n (%) | <.001 | |||
| Yes | 8026 (33.16) | 6254 (28.90) | 1772 (60.94) | |
| No | 13,683 (66.84) | 12,775 (71.10) | 908 (39.06) | |
| Hyperlipidemia, n (%) | <.001 | |||
| Yes | 15,160 (68.94) | 12,969 (66.92) | 2191 (82.10) | |
| No | 6549 (31.06) | 6060 (33.08) | 489 (17.90) | |
| Diabetes, n (%) | <.001 | |||
| Yes | 3172 (10.89) | 2482 (9.36) | 690 (20.90) | |
| No | 18,537 (89.11) | 16,547 (90.64) | 1990 (79.10) | |
| BMI, mean (SE) | 28.63 ± 0.08 | 28.32 ± 0.09 | 30.69 ± 0.22 | <.001 |
| WC, mean (SE) | 98.27 ± 0.22 | 97.33 ± 0.22 | 104.44 ± 0.46 | <.001 |
Continuous measurement data were reported as mean ± SD, and categorical data were described as percentages.
AA = associate degree, BMI = body mass index, DBP = diastolic blood pressure, GED = general educational development, OA = osteoarthritis, PIR = poverty income ratio, SBP = systolic blood pressure, WC = waist circumference.
3.2. The relationship between OA and BMI, WC
Multivariate analysis revealed that both BMI and WC are significantly associated with the incidence of OA. Each 1 kg/m² increase in BMI was associated with a 5%–7% higher prevalence of OA (OR = 1.06, 95% CI: 1.05–1.07); each 1 cm increase in WC was associated with a 2%–3% higher OA prevalence (OR = 1.02, 95% CI: 1.02–1.03). Analysis stratified by BMI clinical categories showed that, compared with the BMI < 18.5 kg/m² group, the OR for prevalent OA in the 25–29.9 kg/m² group was 1.69–1.71 (P = .022–0.036), and in the ≥30 kg/m² group, the OR reached 2.53–2.99 (all P < .001), with a significant dose–response trend (P for trend < .001). The difference between the BMI 18.5–24.9 kg/m² group and the reference group was not statistically significant (P > .05). WC stratified analysis showed that, compared with the normal group, the OR for prevalent OA in the abdominal obesity group was 2.73 (95% CI: 2.41–3.08) in the unadjusted model (model 1), 1.77 (95% CI: 1.54–2.03) after adjusting for demographic factors (model 2), and 1.57 (95% CI: 1.35–1.82) after further adjustment for lifestyle and metabolic factors (model 3); all P < .001. These associations remained stable across different adjusted models (Table 2).
Table 2.
Multivariate logistic regression analyses of OA and BMI, WC.
| Model 1 | Model 2 | Model 3 | ||||
|---|---|---|---|---|---|---|
| OR (95% CI) | P value | OR (95% CI) | P value | OR (95% CI) | P value | |
| BMI (kg/m2) | ||||||
| Continuous | 1.05 (1.04–1.06) | <.001 | 1.07 (1.06–1.08) | <.001 | 1.06 (1.05–1.07) | <.001 |
| Clinical cutoffs | ||||||
| <18.5 | 1.00 (Reference) | 1.00 (Reference) | 1.00 (Reference) | |||
| 18.5–24.9 | 1.11 (0.71–1.74) | .656 | 1.17 (0.72–1.90) | .518 | 1.19 (0.73–1.94) | .481 |
| 25–29.9 | 1.69 (1.08–2.62) | .022 | 1.71 (1.07–2.74) | .027 | 1.69 (1.04–2.73) | .036 |
| ≥30 | 2.53 (1.63–3.92) | <.001 | 2.99 (1.88–4.77) | <.001 | 2.72 (1.67–4.43) | <.001 |
| P for trend | <.001 | <.001 | <.001 | |||
| WC (cm) | ||||||
| Continuous | 1.03 (1.02–1.03) | <.001 | 1.03 (1.02–1.03) | <.001 | 1.02 (1.02–1.03) | <.001 |
| Clinical cutoffs | ||||||
| Normal | 1.00 (Reference) | 1.00 (Reference) | 1.00 (Reference) | |||
| Abdominal obese | 2.73 (2.41–3.08) | <.001 | 1.77 (1.54–2.03) | <.001 | 1.57 (1.35–1.82) | <.001 |
| Model 1: unadjusted for any covariates | ||||||
| Model 2: adjust: gender, race, age | ||||||
| Model 3: adjust: adjust: gender, race, education, alcohol, hypertension, smoke history, hyperlimia, DM, age, marital, PIR | ||||||
BMI = body mass index, CI = confidence interval, DM = diabetes mellitus, OR = odds ratio, PIR = poverty income ratio, WC = waist circumference.
3.3. RCS curve plotting and threshold effect analysis
To explore the association between BMI, WC, and OA, RCS curves and threshold effect analyses showed (Figs. 2 and 3, Tables 3 and 4) that both exhibited a linear positive relationship with OA risk (overall P < .001, test for nonlinearity P > .05). No significant threshold effect was observed (likelihood ratio test for logarithmic interaction P > .05). When BMI was ≥24.79 kg/m² or WC was ≥89 cm, each 1-unit increase was associated with a 6% (OR = 1.06, 95% CI: 1.06–1.07) and 3% (OR = 1.03, 95% CI: 1.02–1.03) higher risk of OA, respectively.
Figure 2.
The association between BMI and osteoarthritis. BMI = body mass index, CI = confidence interval, OR = odds ratio.
Figure 3.
The association between WC and osteoarthritis. CI = confidence interval, OR = odds ratio, WC = waist circumference.
Table 3.
Analysis of threshold effect.
| Outcome | Effect | P value |
|---|---|---|
| Model 1: Fitting model by standard linear regression | 1.06 (1.06–1.07) | <.001 |
| Model 2: Fitting model by 2-piecewise linear regression | ||
| Inflection point | 24.79 | |
| <24.79 | 1.05 (1.00–1.11) | .051 |
| ≥24.79 | 1.06 (1.06–1.07) | <.001 |
| P for likelihood test | .690 |
Table 4.
Analysis of threshold effect.
| Outcome | Effect | P value |
|---|---|---|
| Model 1: Fitting model by standard linear regression | 1.03 (1.02–1.03) | <.001 |
| Model 2: Fitting model by 2-piecewise linear regression | ||
| Inflection point | 89 | |
| <89 | 1.02 (1.00–1.04) | .088 |
| ≥89 | 1.03 (1.02–1.03) | <.001 |
| P for likelihood test | .540 |
Using the RCS model, the dose–response relationship between BMI and WC with the risk of OA was analyzed. The results showed that BMI and WC were positively associated with the risk of OA. The OR values increased with higher BMI and WC, with a more pronounced rise observed in the ranges of BMI ≥ 30 kg/m² and WC ≥ 100 cm. The overall associations for both BMI and WC were statistically significant (P for overall < .001), but no significant nonlinear trends were detected (BMI: P = .505, WC: P = .461).
3.4. Subgroup analyses and interaction tests
Subgroup analyses stratified by gender, age, race, education level, marital status, PIR, smoking status, alcohol consumption, diabetes, hypertension, and hyperlipidemia were conducted. The results of these analyses are presented in Figures 4 and 5. The findings indicate that both BMI and WC were significantly associated with the risk of OA across all study participants (BMI: OR = 1.05, 95% CI: 1.04–1.06, P < .001; WC: OR = 1.02, 95% CI: 1.02–1.03, P < .001). The stratified analyses revealed a highly consistent pattern, indicating that the strength of the association between BMI/WC and OA remains consistent across subgroups. Interaction analysis showed significant modifying effects of race (BMI: interaction P = .002, WC: interaction P < .001) and smoking status (BMI: interaction P = .002, WC: interaction P < .001) on the associations between BMI/WC and OA. However, given the multiple testing performed, the clinical significance of these interactions may require further validation.
Figure 4.
The relationship between BMI and OA according to basic features. Except for the stratification component itself, each stratification factor was adjusted for all other variables (gender, age, race, education, marital, PIR, alcohol, hypertension, smoke, hyperlipidemia, and diabetes). BMI = body mass index, CI = confidence interval, OA = osteoarthritis, OR = odds ratio, PIR = poverty income ratio.
Figure 5.
The relationship between WC and OA according to basic features. Except for the stratification component itself, each stratification factor was adjusted for all other variables (gender, age, race, education, marital, PIR, alcohol, hypertension, smoke, hyperlipidemia, and diabetes). CI = confidence interval, OA = osteoarthritis, OR = odds ratio, PIR = poverty income ratio, WC = waist circumference.
In contrast, the interaction P values for gender (BMI: P for interaction = 0.350, WC: P for interaction = 0.982), age (BMI: P for interaction = 0.700, WC: P for interaction = 0.063), education level (BMI: P for interaction = 0.954, WC: P for interaction = 0.937), marital status (BMI: P for interaction = 0.773, WC: P for interaction = 0.785), PIR (BMI: P for interaction = 0.588, WC: P for interaction = 0.738), alcohol consumption (BMI: P for interaction = 0.532, WC: P for interaction = 0.248), hypertension (BMI: P for interaction = 0.532, WC: P for interaction = 0.966), hyperlipidemia (BMI: P for interaction = 0.262, WC: P for interaction = 0.160), and diabetes (BMI: P for interaction = 0.965, WC: P for interaction = 0.464) were all >0.05. These results suggest that the associations between BMI or WC and OA were relatively stable across these subgroups without significant effect modification.
4. Discussion
A cross-sectional analysis of 21,709 participants in this study showed that higher levels of BMI and WC were significantly positively associated with OA. After conducting RCS and threshold effect analyses, it was found that the association became stronger when BMI was ≥24.79 kg/m² or WC was ≥89 cm. These associations remained significant after adjusting for confounding factors such as demographic characteristics, lifestyle factors, and metabolic indicators.
In recent years, a substantial body of research has been dedicated to exploring the relationship between obesity and OA, with results generally indicating a close association. For instance, a study by Xue et al [22] found that an increase in METS-VF (a measure of the metabolic score for visceral fat) is positively correlated with an increased risk of OA. A study in Canada indicated that BMI is strongly associated with OA involving the knee (regardless of whether other joints such as the hip are also affected), and the higher the degree of obesity, the stronger this association.[23] A study from the United States indicated that an increase in WC elevates the risk of physical function decline in the subsequent year among individuals with knee OA or those at high risk for the condition. Conversely, maintaining a stable WC may help reduce this risk.[24] This study employed a combined analysis of BMI and WC to more systematically assess the association between obesity phenotypes and OA. This strategy not only further confirms the association between obesity and the prevalence of OA but also highlights the critical importance of fat distribution in disease risk assessment. Furthermore, it provides a more directed theoretical foundation for future research. Although the exact mechanisms by which obesity contributes to the development of OA are not fully understood, current research generally categorizes its pathways into 3 main types: excessive joint mechanical load, systemic chronic inflammation, and metabolic disorders mediated by adipose tissue. Excess body weight increases the mechanical load on lower limb joints, particularly accelerating cartilage degeneration in weight-bearing joints such as the knees and hips. Moreover, abnormal mechanical stress on the joints can trigger mTORC1 signaling, thereby regulating the apoptosis and autophagy processes of articular chondrocytes, promoting the onset and progression of OA.[25,26] On the other hand, visceral adipose tissue can secrete a variety of proinflammatory cytokines (such as interleukin-6 and tumor necrosis factor-α) and adipokines (including leptin and adiponectin), which participate in the pathological processes of synovial inflammation and cartilage degradation. Metabolic disorders of fat can also lead to ectopic lipid deposition and lipotoxicity, reduce the activity of antioxidant enzymes, and induce oxidative stress, thereby promoting the development of OA.[27,28] Notably, even if BMI is within the normal range, visceral fat accumulation can still increase the risk of OA through metabolic abnormalities such as insulin resistance. Abnormalities in adipose tissue can trigger metabolic issues and release inflammatory factors that damage the joints. Moreover, related conditions such as type 2 diabetes also increase OA risk due to metabolic abnormalities, even when BMI is normal. This suggests that metabolic disorders may be an independent pathogenic mechanism.[29,30]
The age and sex stratified analyses in this study showed no significant difference in the associations between BMI or WC and OA risk across sexes (interaction P > .05). This finding is consistent with research from the UK Biobank data,[31] which also indicated no significant sex difference in the effect of BMI on OA, and this association was most pronounced in knee OA. Genetic factors may be one of the reasons for the lack of gender difference in the association between BMI and OA. The genetic correlation between BMI and OA shows minimal differences between males and females, indicating similarity in the underlying genetic basis in both sexes. As a measure of obesity, BMI cannot accurately reflect gender differences in fat distribution and metabolism, and is influenced by factors such as age and hormones, which may mask the true differences between the sexes. Another study on carpometacarpal OA also showed that although the association between obesity and OA was stronger in men (hazard ratio: 3.57 vs 1.98), the sex interaction did not reach statistical significance.[32] Regarding age stratification, although the association between BMI and OA risk tended to increase with age (<40 years: OR = 1.03; 40–60 years: OR = 1.05; >60 years: OR = 1.06), the differences across age groups were not statistically significant (P = .700). The association between WC and OA remained relatively stable across age groups (OR ≈ 1.02 in each group), although a trend toward slightly higher risk was observed in individuals over 60 years (interaction P = .063). Based on these findings, it is recommended that in clinical practice, older adults should prioritize controlling BMI while also monitoring changes in WC; younger and middle-aged individuals should maintain a healthy weight and WC over the long term. These results highlight the importance of implementing age-differentiated intervention strategies in OA prevention, while suggesting that obesity management strategies can adopt a unified standard across sexes.
This study also observed interaction effects of race and smoking status on the relationship between obesity and OA, which has significant implications for precision intervention. For example, the association between BMI/WC and OA risk was more pronounced among other Hispanic and other racial groups. This may be related to the relatively higher obesity risk observed in Hispanic populations.[33] Research indicates that Hispanic adults and ethnic minority groups have higher rates of obesity compared with non-Hispanic white adults. Factors contributing to this disparity include living in obesogenic community environments, lack of equitable access to obesity treatment resources, and difficulty accessing affordable healthy food options. Lower socioeconomic status further exacerbates this risk,[34] which may strengthen the association between obesity and OA. Based on these findings, it is recommended that Hispanic populations enhance their awareness of obesity-related health risks, actively participate in weight management, and engage in early interventions for OA to reduce the risk of disease and improve quality of life. In addition, smoking status may play a moderating role in the relationship between obesity and OA risk. Smoking may attenuate the association between obesity indicators and OA; some studies suggest that smoking can lead to weight loss (particularly BMI) by suppressing appetite or increasing metabolic rate, thereby potentially partially weakening the positive association between obesity and OA.[35,36] Based on cohort data from the Korean population, some studies suggest that smoking-induced BMI reduction may be a potential explanatory factor for the association between smoking and a decreased risk of OA.[37] Given these heterogeneous results, public health interventions should consider cultural adaptability and individual differences, with a focus on enhancing health education and weight management for high-risk populations (such as smokers and specific racial/ethnic groups).
This study has several strengths, including a large sample size and strong data representativeness. However, it also has some limitations. First, due to the cross-sectional design, causality cannot be established, and reverse causation may exist; for example, OA could lead to reduced physical activity, thereby increasing the risk of obesity. Second, OA diagnosis was based on self-reporting, which, although validated by previous studies, may still introduce misclassification bias. Furthermore, the exclusion of 7571 participants with missing covariates may have introduced selection bias: baseline characteristic analysis showed that these excluded participants were older (46.38 vs 45.61 years), had higher obesity levels (BMI: 29.33 vs 28.63 kg/m², WC: 99.81 vs 98.27 cm), and had a higher prevalence of metabolic diseases, which may have led the current study to underestimate the true strength of the association between obesity and OA. In addition, although this study included as many relevant covariates as possible, it could not entirely eliminate the influence of other potential covariates, which might introduce some bias into the results. Future research should adopt a prospective cohort design, combined with imaging or clinical diagnoses, to further clarify the causal relationship between obesity and OA. It should also explore the independent roles of interventions such as exercise and anti-inflammatory diets in the primary prevention of OA.
5. Conclusion
This study indicates that obesity and abdominal obesity are significantly positively associated with the prevalence of OA. It is recommended that physicians incorporate WC assessment into routine weight management as an important supplementary indicator for OA risk screening. Through multidisciplinary collaboration and early intervention measures, it is hoped that the burden of OA can be alleviated, patient quality of life improved, and healthcare costs reduced.
Acknowledgments
The authors thank the staff and the participants of the National Health and Nutrition Examination Survey study for their valuable contributions.
Author contributions
Writing – review and editing: Enze Li, Xiaoqing Ding.
Data curation: Tangzheng Nie.
Visualization: Zelin Liu.
Validation: Zhigang Li.
Supervision: Jie Zhang.
Supplementary Material
Abbreviations:
- BMI
- body mass index
- CDC
- Centers for Disease Control and Prevention
- CIs
- confidence interval
- DBP
- diastolic blood pressure
- NCHS
- National Center for Health Statistics
- NHANES
- National Health and Nutrition Examination Survey
- OA
- osteoarthritis
- OR
- odds ratio
- PIR
- poverty income ratio
- RCS
- restricted cubic spline
- SBP
- systolic blood pressure
- WC
- waist circumference.
All participants signed written informed consent.
NHANES is conducted by the Centers for Disease Control and Prevention and the National Center for Health Statistics (NCHS). The NCHS Research Ethics Review Committee reviewed and approved the NHANES study protocol.
The authors have no funding and conflicts of interest to disclose.
The datasets generated during and/or analyzed during the current study are publicly available.
Supplemental Digital Content is available for this article.
How to cite this article: Li E, Ding X, Nie T, Liu Z, Li Z, Zhang J. The relationship between obesity/abdominal obesity and osteoarthritis in American adults: Evidence from NHANES 2005 to 2018. Medicine 2025;104:37(e44539).
EL and XD contributed to this article equally.
Contributor Information
Enze Li, Email: lizhigang.2005@163.com.
Xiaoqing Ding, Email: dxq418207628@163.com.
Tangzheng Nie, Email: 2122953711@qq.com.
Zelin Liu, Email: liuzelin20201007@163.com.
Zhigang Li, Email: lizhigang.2005@163.com.
References
- [1].Molnar V, Matišić V, Kodvanj I, et al. Cytokines and chemokines involved in osteoarthritis pathogenesis. Int J Mol Sci. 2021;22:9208. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [2].Hawker GA, King LK. The burden of osteoarthritis in older adults. Clin Geriatr Med. 2022;38:181–92. [DOI] [PubMed] [Google Scholar]
- [3].Woolf AD, Pfleger B. Burden of major musculoskeletal conditions. Bull World Health Organ. 2003;81:646–56. [PMC free article] [PubMed] [Google Scholar]
- [4].Bruyère O, Cooper C, Arden N, et al. Can we identify patients with high risk of osteoarthritis progression who will respond to treatment? A focus on epidemiology and phenotype of osteoarthritis. Drugs Aging. 2015;32:179–87. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [5].Hales CM, Carroll MD, Fryar CD, et al. Prevalence of obesity and severe obesity among adults: United States, 2017-2018. NCHS Data Brief. 2020;360:1–8. [PubMed] [Google Scholar]
- [6].Hu L, Han X, Chen M, Zhang T. Association of waist circumference and BMI with premature death in young and middle-aged population. Front Public Health. 2024;12:1389766. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [7].Freedman DS, Ford ES. Are the recent secular increases in the waist circumference of adults independent of changes in BMI? Am J Clin Nutr. 2015;101:425–31. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [8].Klein S, Allison DB, Heymsfield SB, et al. Waist circumference and cardiometabolic risk: a consensus statement from shaping America’s Health: Association for Weight Management and Obesity Prevention; NAASO, the Obesity Society; the American Society for Nutrition; and the American Diabetes Association. Obesity (Silver Spring, Md). 2007;15:1061–7. [DOI] [PubMed] [Google Scholar]
- [9].Li Y, Lai C, Pan Z, He Y, Liu C. Abdominal obesity and osteoarthritis risk: evaluating the association of lipid accumulation product, body roundness index, and waist triglyceride index with osteoarthritis in U.S. adults. Front Nutr. 2025;12:1570740. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [10].Duclos M. Osteoarthritis, obesity and type 2 diabetes: the weight of waist circumference. Ann Phys Rehabil Med. 2016;59:157–60. [DOI] [PubMed] [Google Scholar]
- [11].Nan K, Zhang M, Hu S, et al. Relationship of weight change patterns from young to middle adulthood with incident rheumatoid arthritis and osteoarthritis: a retrospective cohort study. Front Endocrinol. 2023;14:1308254. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [12].World Health Organization. Obesity: preventing and managing the global epidemic. Report of a WHO consultation. World Health Organ Tech Rep Ser. 2000;894:i–xii, 1-253. [PubMed] [Google Scholar]
- [13].Koparkar G, Biswas DA. Adiposity and cardiac defects: pathophysiology and etiology. Cureus. 2023;15:e34026. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [14].de Araújo Sugizaki CS, da Silva LLS, de Souza Freitas ATV, et al. Comparison between general obesity and abdominal adiposity to estimate cardiovascular disease prevalence in individuals with chronic kidney disease: results from NHANES 2005-2016. Eur J Clin Nutr. 2024;78:449–51. [DOI] [PubMed] [Google Scholar]
- [15].Huang Y, Zhang X, Li B, et al. Association of BMI and waist circumference with diabetic microvascular complications: a prospective cohort study from the UK Biobank and Mendelian randomization analysis. Diabetes Res Clin Pract. 2023;205:110975. [DOI] [PubMed] [Google Scholar]
- [16].Li Y, Zhu J, Fan J, et al. Associations of urinary levels of phenols and parabens with osteoarthritis among US adults in NHANES 2005-2014. Ecotoxicol Environ Saf. 2020;192:110293. [DOI] [PubMed] [Google Scholar]
- [17].Deng C, Ke X, Lin L, Fan Y, Li C. Association between indicators of visceral lipid accumulation and infertility: a cross-sectional study based on U.S. women. Lipids Health Dis. 2024;23:186. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [18].Yin J, Gong R, Zhang M, et al. Associations between sleep disturbance, inflammatory markers and depressive symptoms: mediation analyses in a large NHANES community sample. Progress Neuro Psychopharmacol Biol Psychiatry. 2023;126:110786. [DOI] [PubMed] [Google Scholar]
- [19].American Diabetes Association. 2. Classification and diagnosis of diabetes: standards of medical care in diabetes-2020. Diabetes Care. 2020;43:S14–s31. [DOI] [PubMed] [Google Scholar]
- [20].Li C, Shang S. Relationship between sleep and hypertension: findings from the NHANES (2007-2014). Int J Environ Res Public Health. 2021;18:7867. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [21].Han Y, Jiang X, Qin Y, Zhao Y, Zhang G, Liu C. A cross-sectional study exploring the relationship between the dietary inflammatory index and hyperlipidemia based on the National Health and Nutrition Examination Survey (2005-2018). Lipids Health Dis. 2023;22:140. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [22].Xue H, Zhang L, Xu J, et al. Association of the visceral fat metabolic score with osteoarthritis risk: a cross-sectional study from NHANES 2009-2018. BMC Public Health. 2024;24:2269. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [23].Badley EM, Zahid S, Wilfong JM, Perruccio AV. Relationship between body mass index and osteoarthritis for single and multisite osteoarthritis of the hand, hip, or knee: findings from a Canadian longitudinal study on aging. Arthritis Care Res. 2022;74:1879–87. [DOI] [PubMed] [Google Scholar]
- [24].Liang H, Si W, Li L, Yang K. Association between body roundness index and osteoarthritis: a cross-sectional analysis of NHANES 2011-2018. Front Nutr. 2024;11:1501722. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [25].Yao Q, Wu X, Tao C, et al. Osteoarthritis: pathogenic signaling pathways and therapeutic targets. Signal Trans Target Ther. 2023;8:56. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [26].Adouni M, Aydelik H, Faisal TR, Hajji R. The effect of body weight on the knee joint biomechanics based on subject-specific finite element-musculoskeletal approach. Sci Rep. 2024;14:13777. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [27].Nedunchezhiyan U, Varughese I, Sun AR, Wu X, Crawford R, Prasadam I. Obesity, inflammation, and immune system in osteoarthritis. Front Immunol. 2022;13:907750. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [28].Fang H, Judd RL. Adiponectin regulation and function. Comprehensive Physiol. 2018;8:1031–63. [DOI] [PubMed] [Google Scholar]
- [29].Binvignat M, Sellam J, Berenbaum F, Felson DT. The role of obesity and adipose tissue dysfunction in osteoarthritis pain. Nat Rev Rheumatol. 2024;20:565–84. [DOI] [PubMed] [Google Scholar]
- [30].Jimenez-Muro M, Soriano-Romani L, Mora G, Ricciardelli D, Nieto JA. The microbiota-metabolic syndrome axis as a promoter of metabolic osteoarthritis. Life Sci. 2023;329:121944. [DOI] [PubMed] [Google Scholar]
- [31].Zhang L, Zhang W, Wu X, et al. A sex- and site-specific relationship between body mass index and osteoarthritis: evidence from observational and genetic analyses. Osteoarthritis Cartilage. 2023;31:819–28. [DOI] [PubMed] [Google Scholar]
- [32].Rydberg M, Dahlin LB, Gottsäter A, Nilsson PM, Melander O, Zimmerman M. High body mass index is associated with increased risk for osteoarthritis of the first carpometacarpal joint during more than 30 years of follow-up. RMD Open. 2020;6:e001368. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [33].Howell CR, Juarez L, Agne AA, et al. Assessing Hispanic/Latino and non-Hispanic white social determinants of obesity among a community sample of residents in the rural Southeast US. J Immigr Minor Health. 2022;24:1469–79. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [34].Williams MS, McKinney SJ, Cheskin LJ. Social and structural determinants of health and social injustices contributing to obesity disparities. Current obesity reports. 2024;13:617–25. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [35].Alruwaili A, King JA, Deighton K, et al. The association of smoking with different eating and dietary behaviours: a cross-sectional analysis of 80 296 United Kingdom adults. Addiction (Abingdon, England). 2024;119:1737–50. [DOI] [PubMed] [Google Scholar]
- [36].Driva S, Korkontzelou A, Tonstad S, Tentolouris N, Katsaounou P. The effect of smoking cessation on body weight and other metabolic parameters with focus on people with type 2 diabetes mellitus. Int J Environ Res Public Health. 2022;19:13222. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [37].Kwon HM, Yang IH, Park KK, Cho B-W, Byun J, Lee W-S. Cigarette smoking and knee osteoarthritis in the elderly: data from the Korean National Health and nutritional examination survey. Exp Gerontol. 2020;133:110873. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.