Skip to main content
NCBI home page
Diabetology & Metabolic Syndrome logoLink to Diabetology & Metabolic Syndrome
. 2025 Dec 17;18:27. doi: 10.1186/s13098-025-02059-y

Unraveling the connection between obesity, metabolic syndrome and osteoarthritis risk: a large-observational study

Junxiang Wang 1,3,#, Leixuan Peng 2,#, Peng Xu 1,3,#, Fanshun Guo 2, Xin Xu 1,3, Jiachen Wang 1,3, Hui Yu 1,3, Ke Xu 1,3, Zhi Yang 1,3, Junfei Guo 1,3,✉, Mingyi Yang 1,3,✉
PMCID: PMC12822255  PMID: 41408645

Abstract

Background

The relationship between metabolic syndrome (MetS) and osteoarthritis (OA) remains debated, necessitating further exploration to clarify potential causal links.

Methods

This study included 271,019 participants from the UK Biobank and NHANES, and conducted Mendelian randomization (MR) analyses to determine the relationship between BMI, MetS, and OA. We used generalized linear modeling and restricted cubic spline plots to identify non-linear associations, as well as a mediation analysis of the possible mediating effect of MetS between BMI and OA. MR analyses were used to assess genetic causality, with sensitivity and subgroup analyses ensuring robustness.

Results

MetS was present in 27.3% of UK Biobank and 32.4% of NHANES participants, with OA prevalence at 17.2% and 12.6%. BMI showed a significant positive association with both MetS and OA, with risk increasing notably at a BMI of 27. MetS mediated a small proportion of the BMI effect on OA but was not an independent risk factor after BMI adjustment [1.03 (P < 0.05) in UK Biobank and 1.18 (P > 0.05) in NHANES]. Waist circumference was the primary MetS component influencing OA risk. Subgroup analyses indicated MetS increased OA risk in younger participants and showed a “U” shaped association across BMI levels. MR results consistent with observational studies.

Conclusions

There’s a complex interplay between BMI, MetS, and OA. Although MetS mediated a small proportion of the BMI effect, it was not an independent risk factor for OA in the overall population. The impact of MetS on the risk of OA varied by age and BMI levels.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13098-025-02059-y.

Keywords: Osteoarthritis, Metabolic syndrome, UK biobank, NHANES, Cross-sectional study, Mendelian randomization

Background

Osteoarthritis (OA), the most prevalent degenerative joint disease, is the leading cause of pain and joint dysfunction globally, affecting over 500 million individuals [1]. Its rising prevalence, driven by an aging population and increasing obesity rates, imposes a significant economic burden on the whole world [2–4]. OA typically affects the hip and knee joints, and cartilage damage, subchondral bone remodeling and synovitis are the main features of it3. Besides aging, genetic predisposition, trauma, and metabolic abnormalities also contribute to OA development [5]. Currently, except for total joint arthroplasty, effective treatments for severe OA are limited. This emphasizes the critical necessity to identify and intervene in potential risk factors to prevent disease onset and progression [1].

Metabolic syndrome (MetS) refers to a cluster of metabolic abnormalities, including centripetal obesity, insulin resistance, dyslipidemia, and hypertension [6]. MetS and its components are linked to a wide range of diseases, including cancer, cardiovascular conditions, and musculoskeletal disorders [7–9]. Obesity is a well-established risk factor for OA, and its frequent coexistence with MetS suggests a plausible connection between MetS and OA [10]. However, the connection between MetS and OA remains unclear and contentious. Some scholars reported that MetS is linked to the onset and progression of OA, particularly in women, and poses an independent risk factor for OA even after controlling for BMI [11, 12]. Conversely, a 32-year longitudinal study involving 6,274 participants found no significant link between MetS components and knee osteoarthritis (KOA) [13]. These conflicting findings, arising from differing research designs and methodologies, urgently require further large-scale investigations to clarify the relationship.

The UK Biobank and NHANES databases, comprising over 500,000 and 100,000 participants respectively, offer robust foundations for extensive population-based research. Mendelian randomization (MR) is an epidemiological method that investigates the causal relationship between exposures and outcomes, utilizing single nucleotide polymorphisms (SNPs) as instrumental variables (IVs). Alleles are allocated randomly during meiosis in accordance with Mendelian principles, forming natural exposure and control groups akin to randomized controlled trials (RCTs). This genetic randomization occurs independently of environmental influences, thereby mitigating confounding factors and reverse causation​ [14].

In this work, we first performed a cross-sectional analysis to systematically examine the associations between BMI, MetS, and OA using data from the UK Biobank and NHANES. Concurrently, we conducted MR analyses utilizing genome-wide association study (GWAS) data to investigate the potential genetic causality between various exposures and OA. By integrating observational data from large populations with genetic evidence from MR studies, we aim to achieve a more reliable and comprehensive understanding of the connection between BMI, MetS, and OA.

Methods and materials

Study population

The UK Biobank is a comprehensive population-based prospective cohort study that has collected phenotypic and genetic data from over 500,000 participants. NHANES is an ongoing biennial survey collecting nutrition and health data from the U.S. population. For this study, we included 264,626 white participants from the UK Biobank and 6,393 non-Hispanic white participants from NHANES after excluding those with missing data, and all participants providing written informed consent.

Ascertainment of exposure and outcomes

BMI data were obtained from the measurements recorded in both the UK Biobank and NHANES databases. The diagnosis of metabolic syndrome was made according to the definition of the International Diabetes Federation, and detailed criteria are displayed in the Supplementary material [15]:

In the UK Biobank, disease diagnoses were coded using the International Classification of Diseases, 10th edition (ICD-10). Diagnoses of hip osteoarthritis (ICD-10: M16) and knee osteoarthritis (ICD-10: M17) were obtained from self-reports, primary care records, and hospital inpatient data. In the NHANES database, diagnoses of OA were obtained from the “Medical Conditions” questionnaire, where participants were asked whether they had received a diagnosis of arthritis from a medical professional and, if so, to specify the type of arthritis. Participants were considered to have OA if they answered “Osteoarthritis” or “Osteoarthritis or degenerative arthritis”.

Covariates

To control for potential confounders, we included the following covariates from both databases: age (years), gender (male or female), income level (average total household income before tax in UK Biobank: < £18,000, £18,000–51,999, > £52,000; family income poverty ratio level in NHANES: < 1.3, 1.3–3.5, > 3.5), education level (below high school, high school or above), drinking habits (no drinking, < 2 times/week, ≥ 2 times/week), smoking status (no smoking, used to smoke, currently smoking), physical activity (MET min/week), uric acid (mg/dL), and serum calcium (mmol/L). Physical activity was quantified using metabolic equivalents (MET), calculated as time × frequency × intensity for all types of activities [16].

Statistical analysis

Observational studies

In the observational component of the study, we first analyzed the relationship between BMI and the risk of MetS and OA using generalized linear models. Four different models were used to assess the effects of various potential confounders by incrementally adding covariates. We then utilized restricted cubic spline (RCS) plots with 4 knots to construct non-linear associations between BMI and the risk of MetS and OA, stratified by gender. After ensuring that the relationship met the criteria for mediation (i.e., a significant relationship between exposure to outcome, exposure to mediator, and mediator to outcome), we included all covariates for mediation analyses to explore the mediating role of MetS in the relationship between BMI and OA. The mediation analysis was conducted under the standard causal assumptions of no unmeasured confounding between exposure, mediator, and outcome variables. To investigate whether MetS was an independent risk factor of OA after adjusting for BMI, we included BMI as a covariate in analysis after adjusting for all other covariates. Meanwhile, four models were used to analyze the impact of the MetS’s different components on OA.

Mendelian randomization studies

The GWAS for MetS was derived from a study of 461,920 individuals of European ancestry by van Walree et al. [17]. To accurately reflect the underlying genetic structure, this study’s results were not corrected for BMI. Data were obtained from the CNCR website.

The GWAS for BMI was sourced from a study by Yengo et al. [18] involving 681,275 individuals of European ancestry, which significantly increased the sample size through meta-analysis. Data were accessed via the GIANT website.

The GWAS data for WC, TG, HDL-C, and FG were obtained from the IEU OpenGWAS project website (https://gwas.mrcieu.ac.uk/) [19, 20]. The GWAS summary data for hypertension were obtained from the FinnGen Database, including 412,113 individuals of European ancestry.

GWAS for OA were sourced from a study by Tachmazidou et al., which performed a genome-wide meta-analysis of approximately 17.5 million single nucleotide variants in up to 455,221 individuals using UK Biobank and arcOGEN resources [21]. The design framework of this study is shown in Fig. 1.

Fig. 1.

Fig. 1

Open in a new tab

Design framework of this research. BMI: Body mass index; GVIF: Generalized variance inflation factor; GWAS: Genome-wide association study; IVW: Inverse variance weighted; MetS: Metabolic syndrome; MR: Mendelian randomization; OA: Osteoarthritis; PSM: Propensity score matching

Mendelian randomization utilized SNPs as IVs, which were chosen to satisfy the following three assumptions: (1) significant association with the exposure, (2) association with the outcome only through the exposure, and (3) no association with confounders affecting the exposure-outcome relationship [22]. In order to find IVs with sufficient validity and satisfying the three basic assumptions of MR, a rigorous process of screening was employed and details are shown in the supplementary material. (1) screening for SNPs significantly associated with exposure at the genome-wide level (P < 5 × 10− 8); (2) the effect of linkage disequilibrium was removed (r2 < 0.001, window size = 10,000 kb): (3) SNPs significantly associated with the outcome (P < 5 × 10− 8) were excluded to avoid pleiotropy of IVs (4) palindromic variants that could not be oriented (i.e., C/G or A/T) were also removed. (5) we calculated the F-values of the IVs, and SNPs with F < 10 were removed to avoid weak instrumental variable bias [23].

We utilized a two-sample univariate Mendelian randomization (UVMR) approach to explore the causal relationship between BMI, MetS, and OA, followed by multivariate Mendelian randomization (MVMR) to explore the genetic causality of BMI and MetS on OA after adjusting for each other, as well as the relationship of MetS components with OA. For UVMR, the inverse-variance weighted (IVW) method was used as the main analysis, with additional analyses conducted using the weighted median, MR-Egger, and weighted mode methods.

Sensitivity analysis and subgroup analysis

In the observational study, we assessed multicollinearity among variables by generalized variance inflation factor (GVIF) analysis, with GVIF < 2 indicating the absence of multicollinearity. After the primary analysis, we performed a 1:1 propensity score matching with BMI at the RCS turning point including all covariates, considering the presence or absence of MetS. We used the nearest neighbor matching method with a 0.2 caliper for PSM, evaluating matching balance through hist and jitter plots [24]. Five models were used to analyze the relationship between BMI, MetS, and OA risk after PSM. Identified key factors influencing the MetS-OA relationship including age, gender, and BMI were then guided subsequent subgroup analyses.

For MR study, we calculated the Cochran’s Q statistic and I² index to assess heterogeneity in the IVW method, and evaluated potential horizontal pleiotropy between IVs by the MR-Egger intercept [25]. MR-PRESSO provides corrected causal effects by removing outliers and was used to detect and correct for horizontal pleiotropy [26]. The symmetry of the funnel plot was also used to assess potential pleiotropy, and the scatterplot assessed the reliability of the results.

Foreign package, rms package, mediation package and mendelR package in R software (version 4.3.1) was used for statistical analyses, and all tests were two-tailed, with p < 0.05 considered to be statistically significant.

Results

Descriptive statistics

This study included a total of 264,626 white adult participants from the UK Biobank and 6,393 from NHANES. The overall study design is shown in Fig. 1, and the baseline characteristics of the participants are provided in Table 1 and Supplementary Table 1. The mean ages of participants were 56.06 ± 8.02 years for the UK Biobank and 50.41 ± 17.82 years for NHANES. The median BMIs were 26.64 [24.09, 29.79] kg/m² and 27.14 [23.85, 31.20] kg/m², respectively. MetS was identified in 27.3% of the UK Biobank participants and 32.4% of NHANES participants. OA was diagnosed in 45,431 UK Biobank participants and 804 NHANES participants, with prevalence rates of 17.2% and 12.6%, respectively. Notably, the mean age and median BMI were higher in participants with OA, and the proportion of women and participants with MetS was also higher compared to non-OA participants.

Table 1.

Baseline characteristics of participants from UK biobank

level Overall (n = 264,626) Non-OA (n = 219,195) OA (n = 45,431) P
Age (mean (SD)) 56.06 (8.02) 55.28 (8.03) 59.81 (6.81) < 0.001
Gender (%)
Male 126,219 (47.7) 105,644 (48.2) 20,575 (45.3) < 0.001
Female 138,407 (52.3) 113,551 (51.8) 24,856 (54.7)
Education (%)
Below high school 125,187 (47.3) 99,351 (45.3) 25,836 (56.9) < 0.001
High School or above 139,439 (52.7) 119,844 (54.7) 19,595 (43.1)
Income_level (%)
Low income 46,536 (17.6) 34,246 (15.6) 12,290 (27.1) < 0.001
Middle income 142,332 (53.8) 116,987 (53.4) 25,345 (55.8)
High income 75,758 (28.6) 67,962 (31.0) 7,796 (17.2)
Alcohol (%)
No drinking 14,279 (5.4) 11,422 (5.2) 2,857 (6.3) < 0.001
Light to moderate drinking 121,353 (45.9) 99,710 (45.5) 21,643 (47.6)
Moderate to heavy drinking 128,994 (48.7) 108,063 (49.3) 20,931 (46.1)
Smoking (%)
No smoking 147,445 (55.7) 124,333 (56.7) 23,112 (50.9) < 0.001
Used to smoke 93,434 (35.3) 74,577 (34.0) 18,857 (41.5)
Now smoking 23,747 (9.0) 20,285 (9.3) 3,462 (7.6)
PA (median [IQR]) 1,668.00 [782.00, 3,360.00] 1,659.00 [777.00, 3,312.00] 1,695.00 [817.00, 3,612.00] < 0.001
BMI (median [IQR]) 26.64 [24.09, 29.79] 26.28 [23.83, 29.23] 28.74 [25.65, 32.60] < 0.001
MetS (%)
No MetS 192,421 (72.7) 164,740 (75.2) 27,681 (60.9) < 0.001
MetS 72,205 (27.3) 54,455 (24.8) 17,750 (39.1)
WC (median [IQR]) 90.00 [80.00, 99.00] 89.00 [80.00, 98.00] 95.00 [85.00, 105.00] < 0.001
HDL-C (median [IQR]) 54.37 [45.55, 65.08] 54.68 [45.82, 65.39] 53.06 [44.63, 63.53] < 0.001
TG (mean (SD)) 153.93 (90.14) 151.51 (89.79) 165.65 (90.89) < 0.001
FG (mean (SD)) 91.72 (20.74) 91.41 (20.71) 93.21 (20.83) < 0.001
SBP (mean (SD)) 139.03 (19.31) 138.48 (19.31) 141.64 (19.14) < 0.001
DBP (mean (SD)) 82.21 (10.61) 82.03 (10.69) 83.07 (10.20) < 0.001
Calcium (mean (SD)) 2.38 (0.09) 2.38 (0.09) 2.38 (0.10) < 0.001
Uric acid (median [IQR]) 5.11 [4.22, 6.08] 5.06 [4.17, 6.03] 5.32 [4.47, 6.32] < 0.001
Open in a new tab

OA: Osteoarthritis; PA: Physical activity; BMI: Body mass index; MetS: Metabolic syndrome; WC: Waist circumference; HDL-C: High density lipoprotein; TG: Triglyceride; FG: Fasting glucose; SBP: Systolic pressure; DBP: Diastolic pressure;

Relationship of BMI to the risk of MetS and OA

We employed four different models to explore the impact of different BMI categories on the risk of MetS and OA, with the results presented in Fig. 2. The analysis demonstrated that BMI was significantly and positively associated with both MetS and OA across all models. Compared to the lowest BMI quartile (Q1), the odds ratios (ORs) increased with higher BMI levels, though the ORs gradually decreased with the inclusion of more covariates. After adjusting for all covariates, the ORs for MetS in the highest BMI quartile (Q4) were 27.78 (26.62–28.99) in the UK Biobank and 21.93 (17.28–27.83) in NHANES. For OA, the ORs in Q4 were 3.71 (3.58–3.84) in the UK Biobank and 1.91 (1.49–2.44) in NHANES.

Fig. 2.

Fig. 2

Open in a new tab

Relationship of BMI to the risk of MetS and OA. (A1): Effect of BMI on MetS risk in UK Biobank; (A2): Effect of BMI on MetS risk in NHANES; (B1): Effect of BMI on OA risk in UK Biobank; (B2): Effect of BMI on OA risk in NHANES; (C1): RCS plot of BMI to MetS in UK Biobank; (C2): RCS plot of BMI to MetS in UK Biobank after sex stratification; (D1): RCS plot of BMI to OA in UK Biobank; (D2): RCS plot of BMI to OA in UK Biobank after sex stratification; (E1): RCS plot of BMI to MetS in NHANES; (E2): RCS plot of BMI to MetS in NHANES after sex stratification; (F1): RCS plot of BMI to OA in NHANES; (F2): RCS plot of BMI to OA in NHANES after sex stratification. BMI: Body mass index; MetS: Metabolic syndrome; OA: Osteoarthritis; model 1: No adjustment; model 2: Adjusted for age and gender; model 3: model 2 + education + income + personal behavior (Smoking, alcohol, physical activity); model 4: model 3 + serum calcium + uric acid; * P < 0.05, ** P < 0.01, and *** P < 0.001

Additionally, restricted cubic spline (RCS) plots revealed non-linear relationships (non-linear P < 0.05) between BMI and outcome events (Fig. 2). The effect of BMI on MetS and OA was minimal at lower BMI levels but increased rapidly beyond a certain threshold. Notably, in the RCS, the turning point for both MetS and OA was around a BMI of 27, where the OR exceeded 1, indicating an elevated risk of these conditions. Age stratification revealed no gender differences in the elevated risk of MetS with increasing BMI, but women exhibited a higher risk of developing OA compared to men at the same BMI level.

Effects of MetS on OA and mediating effects between BMI and OA

We used four different models to explore the association between MetS and the development of OA, with results presented in Fig. 3A. MetS was significantly associated with an increased risk of OA in all analyses from both databases. This correlation was strongest in unadjusted analyses, with ORs of 1.94 (1.90–1.98) in the UK Biobank and 1.72 (1.48–2.00.48.00) in NHANES. After multi-model adjustment, the ORs decreased to 1.55 (1.51–1.59) and 1.36 (1.15–1.60), respectively. Adjusting for age and gender had the strongest impact on this relationship.

Fig. 3.

Fig. 3

Open in a new tab

Analysis of the role of MetS and its components on OA. (A): Results of multi-model analysis of MetS on OA; (B): Mediation analysis of MetS in the relationship between BMI and OA; (C): Multivariate analyses results after mutual adjustment of BMI and MetS; (D1)-(D4): Effect of MetS components on OA risk in UK Biobank; (E1)-(E4): Effect of MetS components on OA risk in NHANES; ACME, Average causal mediation effects (indirect effect); ADE, Average direct effects; BMI: Body mass index; MetS: Metabolic syndrome; OA: Osteoarthritis; model 1: No adjustment; model 2: Adjusted for age and gender; model 3: model 2 + education + income + personal behavior (Smoking, alcohol, physical activity); model 4: model 3 + serum calcium + uric acid; * P < 0.05, ** P < 0.01, and *** P < 0.001.* P < 0.05, ** P < 0.01, and *** P < 0.001

The results of the mediation analysis are displayed in Fig. 3B, which demonstrates that MetS significantly mediates the association between BMI and OA. However, the proportion mediated by MetS was relatively small: 8.6% (7.0%−9.0%) in the UK Biobank and 28.5% (4.1%−75.0%) in NHANES.

Relationship between metabolic abnormalities and OA after adjusting for BMI

To avoid interaction effects between BMI and MetS, they were adjusted for each other as covariates based on the previous model 4, with results shown in Fig. 3C. After adjustment, the effect of BMI to OA remained significant, with ORs of 3.65 (3.51–3.79) in the UK Biobank and 1.75 (1.34–2.28) in NHANES for the Q4 level. However, the OR for MetS decreased substantially, from 1.55 (1.51–1.59) to 1.03 (1.01–1.06) in the UK Biobank, and was no longer statistically significant (OR: 1.18, p > 0.05) in NHANES.

Analysis with four models for five variables of waist circumference (WC), triglyceride (TG), high density lipoprotein cholesterol (HDL-C), diastolic blood pressure (DBP), and fasting glucose (FG) was conducted to explore the connection between individual MetS components and OA after testing for multicollinearity, with results displayed in Fig. 3. Given that SBP exhibits a greater GVIF, we used DBP to represent blood pressure and did not additionally adjust for BMI. WC was the only component that maintained a significant effect on OA after multi-model adjustment, with ORs of 3.54 (3.40–3.69) in the UK Biobank and 1.84 (1.39–2.43) in NHANES. Other metabolic factors, such as DBP and FG, had little effect on OA. In the UK Biobank sample, TG and HDL-C showed a mild protective effect on OA after adjusting for other metabolic factors and covariates, whereas this relationship was not observed in the NHANES sample.

Mendelian randomization analysis

We first performed two-sample UVMR analyses between BMI, MetS, and OA, followed by multivariate Mendelian randomization (MVMR) analyses, with results presented in Fig. 4A. A total of 471, 325, 183, and 450 SNPs were used in UVMR and MVMR, respectively. In UVMR, the OR for BMI on OA was 1.87 (1.76–1.99); the OR for BMI on MetS was 1.88 (1.84–1.92); and the OR for MetS on OA was 1.72 (1.55–1.91). In MVMR, the genetic causal effect of BMI on OA was still significant after mutual adjustment, with an OR of 1.82 (1.61–2.06). However, the effect of MetS on OA was substantially attenuated and no longer significant, with an OR of 1.03 (0.87–1.21).

Fig. 4.

Fig. 4

Open in a new tab

Results of the analysis of univariate MR and multivariate MR. (A): Effects between BMI and MetS and OA in MR analysis; (B): Effects between components of MetS and OA in MR analysis; (C): Results of subgroup analysis of MetS for OA; BMI: Body mass index; CI: Confidence interval; DBP: Diastolic pressure; FG: Fasting glucose; HDL-C: High density lipoprotein; IVW: Inverse variance weighted; MetS: Metabolic syndrome; MVMR: Multivariate Mendelian randomization; OA: Osteoarthritis; OR: Odds ratio; SNP: Single nucleotide polymorphism; TG: Triglyceride; UVMR: Univariate Mendelian randomization; WC: Waist circumference

The results of mediated MR analyses, shown in the Supplementary Fig. 5, indicated that MetS mediated part of the genetic effect of BMI on OA, with a mediator proportion of 34.7% (25.2%−44.2%). The results of these MR analyses are consistent with our observational findings.

To analyze the genetic causal effects of individual metabolic factors on OA, we performed MVMR on different components of MetS, A total of 209 SNPs were used, with results displayed in Fig. 4B. After mutual adjustment, only WC demonstrated a significantly elevated genetic risk for OA, with an OR of 2.26 (1.99–2.58). No significant effects of TG and HDL-C on OA were observed, while hypertension and fasting glucose exhibited weak protective genetic effects on OA.

Subgroup analysis and sensitivity analysis

Subgroup analyses based on age, gender, and BMI revealed that MetS was associated with an elevated risk of OA at younger ages (age < 50), but this association was not significant at older ages (70 ≤ age) (Fig. 4C). Gender-specific analyses showed that the increased risk of OA due to MetS was significant in males (OR: 1.04, 1.00–1.08.00.08) but not in females in the UK Biobank sample. In the NHANES sample, the elevated risk of OA due to MetS was more pronounced in females (OR: 1.29, 1.00–1.65.00.65). Interestingly, MetS exhibited a “U” shaped relationship with OA across different BMI levels from the UK Biobank, with `MetS linked to an increased risk of OA at the lowest (Q1) and highest (Q4) BMI levels, but not at intermediate levels (Q2). Similar trends were also observed in the NHANES sample.

We performed propensity score matching (PSM) on samples from the UK Biobank and NHANES using BMI at the RCS turning point, with or without MetS. Matched baseline information is presented in the Supplementary Tables 8–11. Multi-model analyses, shown in Table 2, indicated that the effect of BMI on OA remained significant across all models, while the effect of MetS on OA diminished and was no longer significant after adjustment, consistent with our previous analyses.

Table 2.

Results of multi-model analysis of BMI and MetS on OA after PSM

UK Biobank model 1 model 2 model 3 model 4 model 5
BMI BMI ≤ 27
BMI > 27 2.04 (1.98–2.10) *** 2.09 (2.03–2.15) *** 2.09 (2.03–2.15) *** 2.08 (2.02–2.14) *** 2.08 (2.02–2.14) ***
MetS Non-MetS
MetS 2.04 (1.98–2.10) *** 1.02 (0.99–1.05) 1.02 (0.99–1.05) 1.01 (0.98–1.04) 0.99 (0.96–1.02)
NHANES model 1 model 2 model 3 model 4 model 5
BMI BMI ≤ 27
BMI > 27 1.26 (1.04–1.53) * 1.24 (1.01–1.52) * 1.25 (1.02–1.53) * 1.25 (1.02–1.53) * 1.24 (1.01–1.53) *
MetS Non-MetS
MetS 1.26 (1.04–1.52) * 1.18 (0.98–1.44) 1.19 (0.98–1.45) 1.20 (0.99–1.46) 1.20 (0.98–1.45)
Open in a new tab

BMI: body mass index; MetS: metabolic syndrome; model 1: No adjustment; model 2: Adjusted for age and gender; model 3: model 2 + education + income + personal behavior (Smoking, alcohol, physical activity); model 4: model 3 + serum calcium + uric acid; model 5: model 4 + * P < 0.05, ** P < 0.01, and *** P < 0.001

For MR analyses, information on instrumental variables used and sensitivity analyses are provided in the Supplementary Table 12. All instrumental variables satisfied F > 10, and no evidence of pleiotropy was found (P > 0.05) in UVMR analyses, with no outliers detected in MR-PRESSO.

Discussion

In this large-scale observational and MR study, we identified non-linear relationships between BMI and the risk of MetS and OA. In both analyses, the risk curves initially flattened and then increased significantly, with a notable turning point around a BMI of 27, aligning with recommendations for anti-obesity treatments​ [27]. Although MetS mediated a small proportion of the effect of BMI on OA, it was not an independent risk factor for OA after adjusting for BMI. Genetic effects of metabolic syndrome may be realised exclusively through BMI rather than an independent way. Among the components of MetS, elevated WC was the primary contributor to OA risk, while other components had limited effects. Subgroup analyses revealed intriguing results: MetS was associated with an increased risk of OA in younger participants even after adjusting for BMI, and a “U” shaped association between MetS and OA risk was observed across different BMI levels. Through the use of multiple methods, our study provides a deeper understanding of the relationship between MetS and OA risk, contributing evidence to the epidemiology, prevention, and management of OA.

Obesity is a well-established risk factor for OA, influencing the condition through increased mechanical loading on joints and causing structural changes that impair joint function, and an increase in waist circumference is typically associated with higher body weight and BMI [28]. There may also be reverse causation between obesity and OA, i.e., patients with OA are less active due to pain and dysfunction, which may exacerbate obesity and thus create a vicious cycle. Controlling and reducing weight is beneficial in preventing and slowing OA progression [29]. The mechanical loading theory posits that OA is primarily driven by the increased weight burden on joints. Furthermore, obesity often leads to lipid abnormalities and other metabolic disruptions, contributing to the rising prevalence of MetS alongside obesity [30]. The increase in body weight is a significant factor linking MetS to the heightened risk of OA. Numerous studies have demonstrated a significant association between MetS and OA prior to adjusting for BMI, but this relationship often disappears after such adjustments, which is consistent with our study [13, 31]. These results suggest that it is most likely the mechanical changes following obesity confounding the association between Mets and OA.

In addition to mechanical loading, inflammation is another crucial factor linking obesity, MetS, and OA. OA is characterized by chronic, low-grade inflammation, encompassing both local and systemic inflammatory processes [32]. Macrophages and cytokines, including IL-1β and TNF, play vital roles in local inflammation, and patients with obesity or MetS often exhibit systemic low-grade inflammation [33]. Previous studies revealed higher levels of pro-inflammatory cytokines were observed in overweight individuals, with adipose tissue macrophages shifting from an anti-inflammatory M2 phenotype to a pro-inflammatory M1 phenotype as weight increases [10]. MetS also influences multiple pathways, such as the AMPK pathway, leading to macrophage polarization towards the pro-inflammatory M1 phenotype [33]. Moreover, hyperlipidemia can also elevate levels of reactive oxygen species and cytokines like IL-6 and IL-8, exacerbating inflammatory responses [34].

Except for abdominal obesity, the connection between other MetS components and OA is complex and controversial. In this study, TG and HDL-C showed a negative correlation to OA in the UK Biobank data and the opposite in NHANES. Various cross-sectional and cohort studies have found associations between hyperlipidemia and increased OA risk [35, 36]. However, Schwager et al. found no significant relationship between dyslipidemia and OA, which is supported by evidence from MR studies [37]. Hypertension may contribute to OA through multiple mechanisms, including angiogenesis, shared molecular pathways, and other factors leading to disease onset and progression [38]. However, a study of 2,234 participants from South Korea found a negative correlation between hypertension and OA after adjusting for multiple covariates [39]. Evidence from MR studies on this relationship is also inconsistent [40, 41]. The association between elevated fasting glucose and OA similarly varies, with different studies yielding conflicting results [42, 43]. These inconsistencies suggest that MetS components other than WC may not play an important role in OA development.

A growing number of studies have shown that classifying obesity solely on the basis of BMI masks heterogeneity in fat distribution and metabolic status. So-called “metabolically healthy obese (MHO)” individuals are at lower risk for traditional cardiovascular outcomes, but are not completely harmless [44]. In contrast, “metabolically unhealthy non-obese (MUNO)” individuals, despite being of normal weight, are at substantial risk for a wide range of chronic diseases, including musculoskeletal disorders [45], due to abdominal adiposity and metabolic abnormalities. Our study showed that the association between metabolic syndrome and osteoarthritis was significantly attenuated and no longer statistically significant after adjusting for BMI, suggesting that metabolic abnormalities alone (without overweight) may have a small effect on osteoarthritis risk in the general population. However, subgroup analyses showed that the effect of metabolic abnormalities was more pronounced in the younger age group and in the lowest BMI quartile, a pattern consistent with the MUNO phenotypic profile.

One key finding from our study is the differential effect of MetS on OA across various age groups. As a degenerative disease, age is a major determinant of OA risk. Järvholm et al.‘s study identified a non-linear relationship between age and KOA, with a sharp increase in OA incidence among men aged 50–70 [46]. This implies that metabolic changes and systemic inflammation due to MetS may be more impactful in younger individuals, whereas joint wear and degeneration become predominant factors as age increases.

The “U” shaped association between MetS and OA risk across different BMI levels is particularly intriguing. Abdominal obesity, a component of MetS, signifies primary metabolic disturbances in patients with low BMI, indicating that metabolic disorders rather than mechanical loading are primary OA risk factors in these individuals. In participants with mid-range BMI, the risk of OA is minimized, possibly due to secondary metabolic abnormalities induced by overweight, where mechanical loading remains a significant joint health risk [47]. At higher BMI levels, MetS represents severe obesity and more extensive metabolic dysfunctions, contributing to the increased OA risk through multiple systemic impacts, including non-MetS-related organ anomalies [48].

Our study has several strengths. Firstly, utilizing large datasets from the UK Biobank and NHANES enhances the study’s credibility. Secondly, we employed multiple analytical methods and models to comprehensively examine the relationships between BMI, MetS, and OA, combining MR with observational studies for mutual validation. Thirdly, extensive sensitivity analyses further ensured the robustness of our findings. However, there still have some limitations. Firstly, both observational and MR studies’ sample populations were primarily white individuals of European ancestry, and the results could not be directly applied to other ancestry populations. Secondly, as a cross-sectional study, causal inference strength is inherently limited, however, we supplemented the MR analysis to enhance the credibility of causality. Thirdly, given that MR reveals the cumulative effect of genetic variants over an individual’s entire lifetime, some of the results may differ from clinical observations. In addition, the two databases have slightly different diagnoses for patients with OA, which may lead to limited comparability between them. Although sensitivity analyses were conducted, the possibility of unknown horizontal pleiotropy cannot be completely excluded, and sample overlap may affect the precision of the estimates. Our findings highlight the distinct associations between metabolic factors and OA across different age groups. Future follow-up and prospective studies may further clarify these changing patterns.

Conclusion

This large observational and MR study demonstrated a non-linear effect of BMI on MetS and OA. MetS mediated part of the BMI effect on OA, but its proportion was low, and metabolic factors were less influential than mechanical factors, age, and sex in OA development. However, metabolic abnormalities increased OA risk in younger individuals and those with lower or very high BMI. This study provides comprehensive epidemiological evidence elucidating the MetS-OA relationship and identifies varying MetS impacts across different populations, offering new insights for future research and management strategies.

Supplementary Information

Supplementary Material 1. (32KB, docx)
Supplementary Material 2. (38.5KB, docx)
Supplementary Material 3. (84KB, xlsx)
Supplementary Material 4. (1.6MB, docx)

Acknowledgements

Thanks to the UK Biobank and NHANES and providers and researchers of all the publicly available datasets used in the study.

Abbreviations

BMI

Body mass index

DBP

Diastolic blood pressure

FG

Fasting glucose

GVIF

Generalized variance inflation factor

GWAS

genome-wide association study

HDL-C

High density lipoprotein cholesterol

ICD-10

International Classification of Diseases, 10th edition

IV

Instrumental variable

IVW

Inverse-variance weighted

MetS

Metabolic syndrome

MR

Mendelian randomization

MVMR

Multivariate Mendelian randomization

OA

Osteoarthritis

OR

Odds ratio

PSM

Propensity score matching

RCS

Restricted cubic spline

RCT

Randomized controlled trial

SNP

Single nucleotide polymorphism

TG

Triglyceride

UVMR

Univariate Mendelian randomization

WC

Waist circumference

Author contributions

All authors made significant contributions to the design and implementation of this study. J.X.W., L.P., M.Y. and J.G. conceived the study. J.X.W. and L.P. performed the data analysis, J.X.W. wrote the first draft of the manuscript. X.X., P.X., and F.G supervised all the analyses and are responsible for the interpretation of the data. Z.Y., J.C.W. and K.X. performed the validation analyses. H.Y., M.Y. and P.X. performed data organization. All authors reviewed the draft manuscript.

Funding

This research was funded by Scientific Research and Innovation Platform for Intelligent and Precise Treatment of Bone and Joint Diseases in Shaanxi Province (No. 2024PT-13), the Science and Technology Program of Xi’an, Shaanxi Province (No. 24YXYJ0085), the 2024 Xi’an Honghui Hospital New Technology and New Business Project, and Postdoctoral Fund of Shaanxi Province (No. 2023BSHGZZHQYXMZZ02).

Data availability

The data supporting the findings from this study are available within the manuscript. This study was conducted using the UK Biobank Resource (Application 46478). Researchers can apply to use the UK Biobank resource for health-related research that is in the public interest ([https://www.ukbiobank.ac.uk/register-apply/]).

Declarations

Ethics approval and consent to participate

This study was based on publicly available databases, and all participants had provided written informed consent in the original studies; therefore, no additional ethical approval was required.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.

Footnotes

Publisher’s note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Junxiang Wang, Leixuan Peng and Peng Xu are co-first authors of the article and contributed equally to this work.

Contributor Information

Junfei Guo, Email: drjfguo@163.com.

Mingyi Yang, Email: ymy25808@163.com.

Reference

  • 1.Katz JN, Arant KR, Loeser RF. Diagnosis and treatment of hip and knee osteoarthritis: a review. JAMA. 2021;325(6):568–78. 10.1001/jama.2020.22171. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2.Long H, Liu Q, Yin H, et al. Prevalence trends of site-specific osteoarthritis from 1990 to 2019: findings from the Global Burden of Disease Study 2019. Arthritis Rheumatol. 2022;74(7):1172–83. 10.1002/art.42089. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.Hunter DJ, March L, Chew M. Osteoarthritis in 2020 and beyond: a lancet commission. Lancet. 2020;396(10264):1711–2. 10.1016/S0140-6736(20)32230-3. [DOI] [PubMed] [Google Scholar]
  • 4.Leifer VP, Katz JN, Losina E. The burden of OA-health services and economics. Osteoarthritis Cartilage. 2022;30(1):10–6. 10.1016/j.joca.2021.05.007. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Gao YH, Zhao CW, Liu B, et al. An update on the association between metabolic syndrome and osteoarthritis and on the potential role of leptin in osteoarthritis. Cytokine. 2020;129:155043. 10.1016/j.cyto.2020.155043. [DOI] [PubMed] [Google Scholar]
  • 6.Fahed G, Aoun L, Bou Zerdan M, et al. Metabolic syndrome: updates on pathophysiology and management in 2021. Int J Mol Sci. 2022;23(2):786. 10.3390/ijms23020786. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Silveira Rossi JL, Barbalho SM, de Reverete Araujo R, Bechara MD, Sloan KP, Sloan LA. Metabolic syndrome and cardiovascular diseases: going beyond traditional risk factors. Diabetes Metab Res Rev. 2022;38(3):e3502. [DOI] [PubMed] [Google Scholar]
  • 8.Fang S, Liu Y, Dai H, et al. Association of metabolic syndrome and the risk of bladder cancer: a prospective cohort study. Front Oncol. 2022;12:996440. 10.3389/fonc.2022.996440. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Huang W, Deng S, Liu S, et al. Association of metabolic syndrome and sarcopenia with all-cause and cardiovascular mortality: a prospective cohort study based on the NHANES. Front Endocrinol (Lausanne). 2024;15:1346669. 10.3389/fendo.2024.1346669. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Sampath SJP, Venkatesan V, Ghosh S, Kotikalapudi N. Obesity, metabolic syndrome, and osteoarthritis—an updated review. Curr Obes Rep. 2023;12(3):308–31. 10.1007/s13679-023-00520-5. [DOI] [PubMed] [Google Scholar]
  • 11.Lee BJ, Yang S, Kwon S, Choi KH, Kim W. Association between metabolic syndrome and knee osteoarthritis: a cross-sectional nationwide survey study. J Rehabil Med. 2019;51(6):464–70. 10.2340/16501977-2561. [DOI] [PubMed] [Google Scholar]
  • 12.Zhang S, Wang D, Zhao J, et al. Metabolic syndrome increases osteoarthritis risk: findings from the UK biobank prospective cohort study. BMC Public Health. 2024;24(1):233. 10.1186/s12889-024-17682-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Konstari S, Sääksjärvi K, Heliövaara M, et al. Associations of metabolic syndrome and its components with the risk of incident knee osteoarthritis leading to hospitalization: a 32-year follow-up study. Cartilage. 2021;13(1_suppl):1445S-1456S. 10.1177/1947603519894731. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Larsson SC. Mendelian randomization as a tool for causal inference in human nutrition and metabolism. Curr Opin Lipidol. 2021;32(1):1–8. 10.1097/MOL.0000000000000721. [DOI] [PubMed] [Google Scholar]
  • 15.Alberti KGMM, Eckel RH, Grundy SM, et al. Harmonizing the metabolic syndrome: a joint interim statement of the international diabetes federation task force on epidemiology and prevention; National heart, lung, and blood Institute; American heart association; world heart federation; international atherosclerosis Society; and international association for the study of obesity. Circulation. 2009;120(16):1640–5. 10.1161/CIRCULATIONAHA.109.192644. [DOI] [PubMed] [Google Scholar]
  • 16.Niu Z, Duan Z, He W, et al. Kidney function decline mediates the adverse effects of per- and poly-fluoroalkyl substances (PFAS) on uric acid levels and hyperuricemia risk. J Hazard Mater. 2024;471:134312. 10.1016/j.jhazmat.2024.134312. [DOI] [PubMed] [Google Scholar]
  • 17.van Walree ES, Jansen IE, Bell NY, et al. Disentangling genetic risks for metabolic syndrome. Diabetes. 2022;71(11):2447–57. 10.2337/db22-0478. [DOI] [PubMed] [Google Scholar]
  • 18.Yengo L, Sidorenko J, Kemper KE, et al. Meta-analysis of genome-wide association studies for height and body mass index in ∼700000 individuals of European ancestry. Hum Mol Genet. 2018;27(20):3641–9. 10.1093/hmg/ddy271. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Willer CJ, Schmidt EM, Sengupta S, et al. Discovery and refinement of loci associated with lipid levels. Nat Genet. 2013;45(11):1274–83. 10.1038/ng.2797. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Chen J, Spracklen CN, Marenne G, et al. The trans-ancestral genomic architecture of glycemic traits. Nat Genet. 2021;53(6):840–60. 10.1038/s41588-021-00852-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.Tachmazidou I, Hatzikotoulas K, Southam L, et al. Identification of new therapeutic targets for osteoarthritis through genome-wide analyses of UK biobank data. Nat Genet. 2019;51(2):230–6. 10.1038/s41588-018-0327-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Sekula P, Del Greco MF, Pattaro C, Köttgen A. Mendelian randomization as an approach to assess causality using observational data. J Am Soc Nephrol. 2016;27(11):3253–65. 10.1681/ASN.2016010098. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.Burgess S, Thompson SG. Avoiding bias from weak instruments in Mendelian randomization studies. Int J Epidemiol. 2011;40(3):755–64. 10.1093/ije/dyr036. [DOI] [PubMed] [Google Scholar]
  • 24.Fu EL, Groenwold RHH, Zoccali C, Jager KJ, van Diepen M, Dekker FW. Merits and caveats of propensity scores to adjust for confounding. Nephrol Dial Transplant. 2019;34(10):1629–35. 10.1093/ndt/gfy283. [DOI] [PubMed] [Google Scholar]
  • 25.Bowden J, Davey Smith G, Burgess S. Mendelian randomization with invalid instruments: effect estimation and bias detection through Egger regression. Int J Epidemiol. 2015;44(2):512–25. 10.1093/ije/dyv080. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.Verbanck M, Chen CY, Neale B, Do R. Detection of widespread horizontal pleiotropy in causal relationships inferred from Mendelian randomization between complex traits and diseases. Nat Genet. 2018;50(5):693–8. 10.1038/s41588-018-0099-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Ryan DH. Next generation antiobesity medications: Setmelanotide, Semaglutide, Tirzepatide and bimagrumab: what do they mean for clinical practice? J Obes Metab Syndr. 2021;30(3):196–208. 10.7570/jomes21033. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Godziuk K, Hawker GA. Obesity and body mass index: past and future considerations in osteoarthritis research. Osteoarthritis Cartilage. 2024;32(4):452–9. 10.1016/j.joca.2024.02.003. [DOI] [PubMed] [Google Scholar]
  • 29.Whittaker JL, Runhaar J, Bierma-Zeinstra S, Roos EM. A lifespan approach to osteoarthritis prevention. Osteoarthritis Cartilage. 2021;29(12):1638–53. 10.1016/j.joca.2021.06.015. [DOI] [PubMed] [Google Scholar]
  • 30.Koren D, Taveras EM. Association of sleep disturbances with obesity, insulin resistance and the metabolic syndrome. Metabolism. 2018;84:67–75. 10.1016/j.metabol.2018.04.001. [DOI] [PubMed] [Google Scholar]
  • 31.Walter SS, Wintermeyer E, Klinger C, et al. Association between metabolic syndrome and hip osteoarthritis in middle-aged men and women from the general population. PLoS One. 2020;15(3):e0230185. 10.1371/journal.pone.0230185. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32.De Roover A, Escribano-Núñez A, Monteagudo S, Lories R. Fundamentals of osteoarthritis: inflammatory mediators in osteoarthritis. Osteoarthritis Cartilage. 2023;31(10):1303–11. 10.1016/j.joca.2023.06.005. [DOI] [PubMed] [Google Scholar]
  • 33.Dickson BM, Roelofs AJ, Rochford JJ, Wilson HM, De Bari C. The burden of metabolic syndrome on osteoarthritic joints. Arthritis Res Ther. 2019;21(1):289. 10.1186/s13075-019-2081-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34.Medina-Luna D, Santamaría-Olmedo MG, Zamudio-Cuevas Y, et al. Hyperlipidemic microenvironment conditionates damage mechanisms in human chondrocytes by oxidative stress. Lipids Health Dis. 2017;16(1):114. 10.1186/s12944-017-0510-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35.Tudorachi NB, Totu T, Eva I, et al. Knee osteoarthritis in relation to the risk factors of the metabolic syndrome components and environment of origin. J Clin Med. 2022;11(24):7302. 10.3390/jcm11247302. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36.Baudart P, Louati K, Marcelli C, Berenbaum F, Sellam J. Association between osteoarthritis and dyslipidaemia: a systematic literature review and meta-analysis. RMD Open. 2017;3(2):e000442. 10.1136/rmdopen-2017-000442. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37.Meng H, Jiang L, Song Z, Wang F. Causal associations of circulating lipids with osteoarthritis: a bidirectional Mendelian randomization study. Nutrients. 2022;14(7):1327. 10.3390/nu14071327. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 38.Ching K, Houard X, Berenbaum F, Wen C. Hypertension meets osteoarthritis — revisiting the vascular aetiology hypothesis. Nat Rev Rheumatol. 2021;17(9):533–49. 10.1038/s41584-021-00650-x. [DOI] [PubMed] [Google Scholar]
  • 39.Han CD, Yang IH, Lee WS, Park YJ, Park KK. Correlation between metabolic syndrome and knee osteoarthritis: data from the Korean National health and nutrition examination survey (KNHANES). BMC Public Health. 2013;13:603. 10.1186/1471-2458-13-603. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 40.Gill D, Karhunen V, Malik R, Dichgans M, Sofat N. Cardiometabolic traits mediating the effect of education on osteoarthritis risk: a Mendelian randomization study. Osteoarthritis Cartilage. 2021;29(3):365–71. 10.1016/j.joca.2020.12.015. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 41.Funck-Brentano T, Nethander M, Movérare-Skrtic S, Richette P, Ohlsson C. Causal factors for knee, hip, and hand osteoarthritis: a Mendelian randomization study in the UK biobank. Arthritis Rheumatol. 2019;71(10):1634–41. 10.1002/art.40928. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 42.Li X, Zhou Y, Liu J. Association between prediabetes and osteoarthritis: a meta-analysis. Horm Metab Res. 2022;54(2):104–12. 10.1055/a-1730-5251. [DOI] [PubMed] [Google Scholar]
  • 43.Zheng J, Huang X, Huang J, et al. Association of diabetes mellitus status and hyperglycemia with symptomatic knee osteoarthritis. Arthritis Care Res (Hoboken). 2023;75(3):509–18. 10.1002/acr.24872. [DOI] [PubMed] [Google Scholar]
  • 44.Fauchier G, Bisson A, Bodin A, et al. Metabolically healthy obesity and cardiovascular events: a nationwide cohort study. Diabetes Obes Metab. 2021;23(11):2492–501. 10.1111/dom.14492. [DOI] [PubMed] [Google Scholar]
  • 45.Karimi S, Pasdar Y, Hamzeh B, Ayenehpour A, Heydarpour F, Goudarzi F. Obesity phenotypes related to musculoskeletal disorders; a cross-sectional study from RaNCD cohort. Arch Public Health. 2022;80(1):185. 10.1186/s13690-022-00947-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 46.Järvholm B, Lewold S, Malchau H, Vingård E. Age, bodyweight, smoking habits and the risk of severe osteoarthritis in the hip and knee in men. Eur J Epidemiol. 2005;20(6):537–42. 10.1007/s10654-005-4263-x. [DOI] [PubMed] [Google Scholar]
  • 47.Piché ME, Tchernof A, Després JP. Obesity phenotypes, diabetes, and cardiovascular diseases. Circ Res. 2020;126(11):1477–500. 10.1161/CIRCRESAHA.120.316101. [DOI] [PubMed] [Google Scholar]
  • 48.Schetz M, De Jong A, Deane AM, et al. Obesity in the critically ill: a narrative review. Intensive Care Med. 2019;45(6):757–69. 10.1007/s00134-019-05594-1. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary Material 1. (32KB, docx)
Supplementary Material 2. (38.5KB, docx)
Supplementary Material 3. (84KB, xlsx)
Supplementary Material 4. (1.6MB, docx)

Data Availability Statement

The data supporting the findings from this study are available within the manuscript. This study was conducted using the UK Biobank Resource (Application 46478). Researchers can apply to use the UK Biobank resource for health-related research that is in the public interest ([https://www.ukbiobank.ac.uk/register-apply/]).

Articles from Diabetology & Metabolic Syndrome are provided here courtesy of BMC

RESOURCES