Global burden of high BMI-related osteoarthritis and causal effects of obesity: a GBD 1990–2021 and bidirectional two-sample Mendelian randomization study

Abstract

Background:

Osteoarthritis is the most prevalent joint disorder, and osteoarthritis associated with high BMI contributes significantly to the global disease burden. The aim of this study is to elucidate the burden of BMI-related osteoarthritis and to investigate the causal relationship between varying degrees of obesity and osteoarthritis.

Methods:

The Global Burden of Disease (GBD) 2021 study was used to conduct an epidemiological analysis of osteoarthritis. This included descriptive statistics, age–sex differences, Joinpoint regression analysis, burden sources and control measures, and projections up to 2050. A two-sample Mendelian randomization analysis was also performed to explore the relationship between obesity and osteoarthritis.

Results:

Osteoarthritis burden rose from 1990 to 2021, affecting younger populations, though there was a decline in regions with a high Socio-demographic Index (SDI) between 2001 and 2005. Women experienced a greater burden than men. The increase in disease burden was driven by population growth and rising disease prevalence. There is considerable potential for enhancing osteoarthritis control in most countries. Forward Mendelian randomization analyses demonstrated a significant causal relationship between obesity and osteoarthritis, whereas inverse Mendelian randomization analyses found no evidence of causality from osteoarthritis to obesity at any obesity level.

Conclusion:

The global rise in disability-adjusted life-years (DALYs) due to BMI-related osteoarthritis mainly due to population growth, affecting women more than men. This study may be a reference point for early and precise intervention in BMI-related osteoarthritis.

Introduction

Osteoarthritis (OA), the most prevalent joint disorder, is a chronic, degenerative, and disabling condition that imposes substantial health, medical, and economic burdens^[1,2]. Body mass index (BMI) is the only recognized risk factor for osteoarthritis in current Global Burden of Disease (GBD)studies, and obesity significantly increases the risk of developing osteoarthritis, thus, the causal relationship between BMI and osteoarthritis has become a critical research focus^[1,3,4]. Comprehensive analyses of the osteoarthritis burden attributable to high BMI are currently available only up to 2019, with limited data provided for 2021^[4-6]. Although some studies have investigated the causal relationship between obesity and osteoarthritis substantial knowledge gaps remain^[7]. Previous research on the osteoarthritis disease burden attributable to BMI extended only to 2019 and primarily focused on correlations with the Socio-demographic index (SDI)^[6]. While Mendelian randomization (MR) analyses have explored the overall causal relationship between BMI and osteoarthritis more comprehensively, they have not adequately examined causality across different obesity grades^[8]. Therefore, updated analyses based on GBD 2021 data and further investigation into the relationship between obesity grading and osteoarthritis are warranted.

HIGHLIGHTS

• A combination of the GBD database and a Mendelian randomization analysis is employed.

• The global disease burden of osteoarthritis attributable to high BMI is increasing both now and in the future.

• The burden of osteoarthritis associated with high BMI is primarily due to population growth and epidemiological change.

• The relationship between different grades of obesity and osteoarthritis appears to be one of unidirectional causality.

To address limitations in existing research on the osteoarthritis burden attributable to high BMI, this study utilized disability-adjusted life-years (DALYs) data from the GBD 2021 database. Descriptive analyses, decomposition analyses, frontier analyses, and Joinpoint regression analyses were performed to evaluate global, regional, and national disease burden trends from 1990 to 2021, with projections up to 2050 generated using Auto Regressive Integrated Moving Average Model (ARIMA) models. Additionally, MR analyses were conducted to investigate the causal relationship between varying degrees of obesity and osteoarthritis, further elucidating the strength of this relationship across different obesity levels. To ensure transparency regarding the use of AI in this manuscript, we have completed the TITAN Guideline checklist in accordance with the requirements of the TITAN Guidelines 2025, which provides detailed information about the AI tools used in this paper and the content involved^[9].

Materials and methods

GBD data sources

Disease burden data utilized in this study were obtained from the GBD 2021 database, covering 204 countries and territories, 371 diseases, and 88 risk factors, each reported with uncertainty intervals^[10]. GBD quantifies development levels for each country or region using the SDI, categorizing them into low, low-middle, middle, high-middle, and high levels. Within GBD 2021, high BMI is defined as a BMI ≥25 kg/m^2[1]. This study extracted data on BMI-related osteoarthritis burden globally, as well as for 204 countries, 21 GBD subregions, and 5 SDI-based regions for individuals aged 30 to 95+ years. The extracted metrics included DALYs and age-standard DALYs rate (ASDR) per 100 000 population.

Instrumental variables acquisition

Instrumental variables (IVs) for obesity and osteoarthritis (including osteoarthritis, osteoarthritis of the hip or knee and other osteoarthritis) were obtained from the IEU Open GWAS database, including genome-wide association study (GWAS) datasets ieu-a-90 (obesity class 1, BMI: 30–35 kg/m²), ieu-a-91 (obesity class 2, BMI: 35–40 kg/m²), and ieu-a-92 (obesity class 3, BMI ≥40 kg/m²)^[11]. The osteoarthritis GWAS datasets utilized were ukb-b-14486 (osteoarthritis), ebi-a-GCST005812 (osteoarthritis of the hip or knee), and ukb-d-M19 (other osteoarthritis). Detailed information on IVs is presented in Supplemental Digital Content Table S1, available at: http://links.lww.com/JS9/E831. The IVs for osteoarthritis of the first carpometacarpal joint were extracted from the UK Biobank (UKB) database using the phenotype code M18. However, the database did not provide information on the number of cases and controls. We extracted SNPs data relating to osteoarthritis of the first carpometacarpal joint from the UKB database. As the exact number of cases and controls was not provided, we screened the data based on the ICD-10 code M18. The final number of included SNPs was 26. MR analysis utilizes Mendel’s laws of inheritance and single nucleotide polymorphisms (SNPs) as IVs to investigate causal relationships between exposures and outcomes. To ensure result validity, MR analysis relies on three critical assumptions regarding the selected SNPs as IVs. Firstly, the IVs must be robustly associated with the exposure of interest (relevance assumption). Secondly, the IVs must not be associated with any known or unknown confounding factors (independence assumption).Thirdly, the IVs affect the outcome exclusively through the exposure and not via any alternative direct causal pathways (exclusion restriction assumption)^[12].

GWAS Data Screening

To ensure adherence to the relevance, independence, and exclusion restriction assumptions required for MR analysis, the following criteria were applied for selecting IVs. Firstly, since fewer than five SNPs met the conventional genome-wide significance threshold (P < 5 × 10⁻⁸) for obesity class 3, we adopted a relaxed threshold (P < 5 × 10⁻⁶) to ensure an adequate number of instrumental variables. This approach also maintained consistent screening criteria across all study groups to enhance comparability^[13,14]. Secondly, linkage disequilibrium (LD) pruning was performed with parameters set at r² < 0.001 and kb >10 000. Finally, IVs were assessed for strength using an F-statistic >10.

To minimize heterogeneity and horizontal pleiotropy, leave-one-out sensitivity analyses and MR pleiotropy residual sum and outlier (MR-PRESSO) tests were conducted, resulting in the exclusion of SNPs rs7164558, rs11847697, and rs2030323 in the obesity class 1 group, and SNPs rs8051591 and rs2112347 in the obesity class 2 group. In the inverse analyses, SNP rs13107325 was excluded from obesity class 1 and obesity class 2 groups.

Data analysis

All statistical analyses in this study were performed using RStudio (version 4.4.2). Joinpoint regression analysis was conducted with Joinpoint software (version 5.3.0), and MR analyses were primarily executed using the TwoSampleMR package (version 0.6.8). All visualizations related to the GBD 2021 data were generated using R. Five methods were applied for MR analyses; the inverse variance weighted (IVW) method served as the primary analytic approach, with the remaining four methods used for supplementary analyses^[15]. A statistical significance threshold of P < 0.05 indicated a causal relationship between exposure and outcome, and an OR >1 indicated a positive association. In this study, the IVW method was used as the main reference for the MR analysis results. To ensure the robustness and reliability of results, tests for heterogeneity and pleiotropy were also performed.

Results

Descriptive analysis

The disease burden of high BMI-related osteoarthritis increased across all GBD regions from 1990 to 2021, though the extent of increase varied by region. From a global perspective (Table 1), the number of DALYs due to high BMI-related osteoarthritis increased from 1 449 681.331 (−127 135.501 to 4 113 313.240) in 1990 to 4 422 953.511 (−421 237.186 to 12 338 124.925) in 2021, with a percentage change of 0.406 (0.344 to 0.511), indicating a rising global burden of high BMI-related osteoarthritis. At the regional level, the highest DALYs in both 1990 and 2021 were observed in Western Europe and East Asia, with 282 127.821 (−25 876.851 to 786 737.657) and 1 201 663.785 (−108 094.928 to 3 403 964.656), respectively. High-income North America recorded the highest ASDR in both years: 67.008 (−6.644 to 178.763) in 1990 and 82.026 (−9.136 to 215.008) in 2021. South Asia showed the greatest percentage increase in the ASDR over the same period, reaching 1.110 (0.904 to 1.356). The distribution of DALYs across GBD regions exhibited substantial heterogeneity in both 1990 and 2021, with East Asia in particular showing a markedly higher DALYs burden in 2021 compared to other regions. From the perspective of SDI subregions, the ASDR was highest in the High SDI group in both 1990 and 2021, whereas this group showed the lowest percentage change among the five SDI categories. Analysis of SDI subregions revealed that while the overall disease burden increased with rising SDI, the rate of increase tended to slow as SDI levels rose. Furthermore, the estimated annual percentage change (EAPC) and the lower bounds of the 95% confidence intervals were greater than zero across all GBD and SDI regions, indicating a consistent upward trend in the global burden of high BMI-related osteoarthritis. At the national level, China and the United States had the highest DALYs in 1990 and 2021, respectively. The United States also consistently reported the highest ASDR globally throughout the study period. To visualize and contrast the burden of disease across countries, we mapped DALYs and ASDR onto a world map (Fig. 1).

Table 1.

The case number and ASDR of DALYs of OA caused by a high BMI in 1990 and 2021 for both sexes by SDI quintiles and by GBD regions, with PC and EAPC of ASDR from 1990 to 2021

Location	1990		2021		PC(95% CI)	EAPCs(95% CI)
	Number(95% UIs)	ASDR(95% UIs)	Number(95% UIs)	ASDR(95% UIs)
Global	1 449 681.331(−127 135.501 to 4 113 313.240)	35.973(−3.137 to 102.312)	4 422 953.511(−421 237.186 to 12 338 124.925)	50.587(−4.805 to 141.349)	0.406(0.344–0.511)	1.17(1.15,1.20)
GBD regions
East Asia	256 084.426(−20 939.602 to 755 795.371)	27.456(−2.229 to 81.484)	1 201 663.785(−108 094.928 to 3 403 964.656)	52.852(−4.726 to 150.444)	0.925(0.747–1.131)	2.47(2.33,2.60)
Oceania	1315.016(−122.573 to 3651.541)	38.936(−3.571 to 109.058)	4313.512(−449.914 to 11 532.698)	48.862(−4.961 to 132.978)	0.255(0.196–0.360)	0.69(0.63,0.76)
Southeast Asia	41 752.593(−3098.966 to 125 492.995)	14.756(−1.086 to 44.765)	201 324.622(−15 981.222 to 582 465.035)	27.872(−2.182 to 80.987)	0.889(0.777–1.061)	2.19(2.13,2.26)
Central Europe	63 283.492(−5875.683 to 172 456.647)	42.046(−3.894 to 114.924)	107 282.008(−11 052.477 to 289 260.699)	51.248(−5.318 to 138.206)	0.219(0.164–0.336)	0.66(0.64,0.67)
Central Asia	16 488.352(−1541.002 to 45 709.183)	34.797(−3.209 to 97.262)	36 183.435(−3859.629 to 97 115.195)	42.044(−4.372 to 113.441)	0.208(0.153–0.321)	0.62(0.61,0.63)
Eastern Europe	120 261.523(−11 249.123 to 330 991.370)	42.777(−3.956 to 118.191)	189 331.508(−19 437.221 to 497 720.308)	54.949(−5.688 to 144.571)	0.285(0.206–0.450)	0.88(0.85,0.90)
High-income Asia Pacific	82 638.795(−6623.006 to 243 996.360)	39.934(−3.197 to 117.825)	201 743.236(−15 797.035 to 603 491.748)	51.496(−4.144 to 153.079)	0.290(0.207–0.377)	0.86(0.84,0.87)
Australasia	13 064.085(−1250.838 to 35 716.432)	56.541(−5.428 to 154.087)	39 024.356(−3963.405 to 105 319.166)	79.242(−8.194 to 213.398)	0.401(0.280–0.657)	1.10(1.04,1.15)
Southern Latin America	26 719.704(−2619.978 to 71 378.730)	57.234(−5.610 to 153.033)	64 937.711(−7109.225 to 167 296.885)	76.306(−8.388 to 196.564)	0.333(0.245–0.508)	0.94(0.88,1.00)
Western Europe	282 127.821(−25 876.851 to 786 737.657)	50.712(−4.686 to 140.959)	525 194.691(−51 633.497 to 1 415 589.915)	62.954(−6.272 to 168.561)	0.241(0.185–0.344)	0.68(0.64,0.72)
High-income North America	223 661.449(−22 013.420 to 599 742.694)	67.008(−6.644 to 178.763)	501 346.752(−55 304.038 to 1 308 805.363)	82.026(−9.136 to 215.008)	0.224(0.141–0.414)	0.47(0.31,0.62)
Caribbean	11 257.846(−978.889 to 31 879.417)	42.821(−3.716 to 121.231)	32 228.652(−3203.638 to 86 894.800)	59.709(−5.930 to 160.976)	0.394(0.312–0.556)	1.14(1.10,1.18)
Central Latin America	44 279.162(−4269.905 to 121 092.827)	50.847(−4.846 to 139.836)	174 200.546(−18 965.680 to 458 370.888)	67.692(−7.324 to 178.752)	0.331(0.251–0.491)	0.91(0.90,0.93)
Andean Latin America	9894.271(−919.012 to 27 077.596)	46.292(−4.258 to 127.243)	39 752.310(−4229.505 to 105 662.703)	65.714(−6.952 to 175.471)	0.420(0.323–0.590)	1.18(1.14,1.21)
North Africa and Middle East	72 032.854(−6988.647 to 198 284.446)	40.230(−3.828 to 111.534)	296 602.757(−34 260.249 to 774 588.430)	59.764(−6.782 to 155.919)	0.486(0.362–0.766)	1.28(1.27,1.29)
South Asia	84 239.633(−6104.066 to 248 489.158)	13.182(−0.940 to 39.289)	439 406.190(−35 032.976 to 1 289 375.790)	27.811(−2.195 to 81.776)	1.110(0.904–1.356)	2.65(2.58,2.73)
Tropical Latin America	43 719.278(−3904.627 to 120 749.452)	45.647(−4.020 to 126.658)	165 101.254(−16 732.597 to 441 786.366)	62.955(−6.350 to 168.570)	0.379(0.294–0.535)	1.07(1.04,1.09)
Central Sub-Saharan Africa	4802.479(−338.394 to 14 209.671)	19.600(−1.371 to 58.773)	21 121.492(−1648.836 to 62 061.115)	34.181(−2.591 to 101.866)	0.744(0.571–0.950)	1.78(1.73,1.83)
Southern Sub-Saharan Africa	12 078.607(−1082.540 to 33 236.083)	43.151(−3.829 to 120.076)	36 137.021(−3684.361 to 95 528.350)	59.335(−5.988 to 158.720)	0.375(0.294–0.556)	1.05(1.03,1.06)
Eastern Sub-Saharan Africa	14 443.587(−1065.742 to 43 728.936)	17.903(−1.311 to 54.972)	54 046.784(−4255.868 to 157 360.215)	28.755(−2.222 to 84.571)	0.606(0.520–0.720)	1.56(1.53,1.58)
Western Sub-Saharan Africa	25 536.359(−2047.467 to 75 155.688)	27.470(−2.175 to 81.735)	92 010.890(−8076.143 to 256 558.550)	42.195(−3.603 to 119.632)	0.536(0.439–0.688)	1.36(1.32,1.41)
SDI quintiles
Low SDI	41 133.194(−3041.863 to 123 324.510)	16.808(−1.228 to 51.226)	150 288.281(−11 967.924 to 440 893.359)	26.731(−2.086 to 79.823)	0.590(0.512–0.684)	1.55(1.51,1.59)
Low-middle SDI	125 910.186(−10 241.838 to 369 632.672)	19.319(−1.549 to 57.407)	527 571.304(−46 776.092 to 1 497 615.068)	34.546(−3.027 to 98.530)	0.788(0.676–0.938)	2.02(1.98,2.06)
Middle SDI	318 093.441(−26 116.873 to 922 529.502)	28.732(−2.329 to 83.675)	1 364 967.962(−126 086.152 to 3 863 690.877)	48.164(−4.413 to 137.153)	0.676(0.564–0.826)	1.86(1.79,1.92)
High-middle SDI	393 635.118(−35 314.978 to 1 102 209.192)	39.000(−3.477 to 109.562)	1 105 143.768(−108 304.992 to 3 016 330.753)	55.846(−5.465 to 152.488)	0.432(0.367–0.544)	1.26(1.22,1.31)
High SDI	569 083.033(−52 686.323 to 1 575 580.469)	52.633(−4.894 to 145.380)	1 271 056.919(−127 983.260 to 3 441 561.020)	66.701(−6.823 to 179.942)	0.267(0.203–0.395)	0.71(0.67,0.76)

Figure 1.

The case number of DALYs and the ASDR of osteoarthritis associated with high BMI. (A) The case number of DALYs of osteoarthritis associated with high BMI in 1990. (B) The case number of DALYs of osteoarthritis associated with high BMI in 2021. (C) The ASDR of osteoarthritis associated with high BMI in 1990. (D) The ASDR of osteoarthritis associated with high BMI in 1990.

Age–sex difference analysis

In both 1990 and 2021, DALYs among females were higher than those among males across all age groups (Supplemental Digital Content Figure S1, available at: http://links.lww.com/JS9/E824). In each age group, female DALYs were approximately 1.6 to 4.1 times higher than those of males. In 1990, the number of DALYs peaked in the 60–64 age group for both sexes, with 145 451 DALYs in females and 86,732 in males. In 2021, the peak shifted to a younger age group (55–59 years), with DALYs reaching 418 154 in females and 258 052 in males, indicating both a shift toward earlier disease burden and a greater increase in burden among females. These findings highlight that the peak age of DALYs of high BMI-related osteoarthritis occurred earlier in 2021 compared to 1990, and that females consistently bore a significantly higher burden than males, with an increasing disparity over time. Additionally, we analyzed crude DALYs rates across age groups. In both 1990 and 2021, the highest crude DALYs rates were observed in the 75–79 age group. However, in 2021, there was a notable trend toward younger age groups, with the crude DALYs rates in the 70–74 and 75–79 age groups becoming nearly equivalent. This suggests that, by 2021, the disease burden in the 70–74 age group had increased to the point of equaling or surpassing that in the older group, further supporting the observation of an earlier age shift in the burden of high BMI-related osteoarthritis.

Joinpoint analysis

Joinpoint regression analysis was employed to identify significant inflection points in time-series data and to segment trends accordingly. We applied Joinpoint analysis to examine temporal trends in the ASDR of high BMI-related osteoarthritis across global, sex-specific, and SDI-region-specific populations. As shown in Figure 2, the ASDR exhibited an overall upward trend globally and across most SDI regions from 1990 to 2021, with the exception of the High-SDI region, which showed a divergent pattern.

Figure 2.

Joinpoint analysis of both sex ASDR of osteoarthritis associated with high BMI. (A) Joinpoint analysis in global. (B) Joinpoint analysis in high SDI region. (C) Joinpoint analysis in high middle SDI region. (D) Joinpoint analysis in middle SDI region. (E) Joinpoint analysis in low middle SDI region. (F) Joinpoint analysis in low SDI region.

In all SDI regions, except the High-SDI group, trends in disease burden did not differ significantly in overall trajectory. Most regions experienced the most rapid increase in ASDR between the second and third Joinpoints, after which the rate of increase slowed. Notably, the Middle-SDI region demonstrated the steepest rate of increase between 2001 and 2004, the second and third Joinpoints during the 1990–2021 period. Furthermore, sex-specific Joinpoint analyses in the High-SDI region revealed distinct trends compared to other SDI regions. Both male and female populations in this region exhibited a declining trend in ASDR between the second and third Joinpoints (Supplemental Digital Content Figure S2, available at: http://links.lww.com/JS9/E825; Supplemental Digital Content Figure S3, available at: http://links.lww.com/JS9/E826), followed by slight fluctuations in the upward trend when compared to other regions.

Analysis of the sources and control of the disease burden

In the frontier analysis, the solid black line represents the theoretical minimum ASDR achievable at various levels of SDI, with each country represented by a dot. The change in ASDR from 1990 to 2021 is indicated by the color of the dots, with red representing a decrease and blue an increase. The absolute distance between a country’s observed ASDR and the frontier is defined as the effective difference, reflecting unrealized health gains relative to the country’s level of development. In 2021, the 15 countries with the greatest deviations from the frontier are labeled in black font. Additionally, five countries with low SDI values (SDI < 0.5) and the smallest deviations are labeled in blue font, while five countries with high SDI values (SDI > 0.85) and the largest deviations are labeled in red font. To further assess the effectiveness of osteoarthritis control efforts related to high BMI, we analyzed the ASDR across all countries from 1990 to 2021. The results indicate that most countries did not reach the minimum achievable burden over this period, suggesting substantial room for improvement. However, many countries did experience a decline in disease burden during this time. As shown in Figure 3, Somalia, Papua New Guinea, Yemen, Timor-Leste, and the Lao People’s Democratic Republic had the smallest effective differences, while the Netherlands, Germany, Canada, Iceland, and the United States had the largest. Somalia and the Netherlands, as the two regions with the lowest and highest effective differences, respectively, exhibit not only significant disparities in disease control potential but also distinct differences in growth trends. Somalia witnessed a decline in disease burden between 1990 and 2021, while the Netherlands demonstrated an opposite trend. Notably, effective differences tended to increase with higher levels of socio-demographic development, indicating that countries with higher SDI values may have greater unrealized potential for reducing the burden of high BMI-related osteoarthritis.

Figure 3.

Analysis of the sources and control of the disease burden. (A, B) Frontier analysis of the relationship between SDI and the burden of osteoarthritis associated with high BMI in 2021. (C) DALYs decomposition analysis of high BMI-related osteoarthritis in both sex across 21 GBD regions and 5 SDI subregions. (D) DALYs decomposition analysis of high BMI-related osteoarthritis in female. (E) DALYs decomposition analysis of high BMI-related osteoarthritis in male.

The combined effects of population growth, aging, and epidemiological changes collectively contributed to an increased burden of high BMI-related osteoarthritis. Across global, GBD regional, and SDI subregional levels, the burden remained consistently higher in females than in males. Overall, population growth was the predominant driver of the global burden, accounting for 56.879%, 58.376%, and 54.729% of the increase in DALYs for the total population, females, and males, respectively (Supplemental Digital Content Table S2, available at: http://links.lww.com/JS9/E832), followed by epidemiological changes, while aging had the least impact. Notably, aging contributed negatively to the disease burden in specific regions, including Western Sub-Saharan Africa, Eastern Sub-Saharan Africa, Central Sub-Saharan Africa, Central Asia, and low SDI regions. Analysis across SDI subregions indicated that as SDI increased, the contribution of population growth to disease burden declined, whereas the impact of aging became more pronounced. Eastern Europe and East Asia differed from other regions in that epidemiological changes were the leading contributors to the disease burden, accounting for 54.887% and 41.984%, respectively. In contrast, most other regions showed a similar burden structure, with population growth serving as the primary driver.

SDI correlation analysis

A scatterplot depicting the relationship between the ASDR and the SDI across 204 countries in 2021 (Fig. 4) revealed a moderate positive correlation (R = 0.576, P < 0.05). The results indicate that the burden of high BMI-related osteoarthritis generally increased with rising SDI, suggesting a positive association between SDI and disease burden. Although a brief decline was observed in high-SDI regions, this was followed by a sharp increase, and overall ASDR remained higher than expected. In particular, Western Europe showed DALYs rates that exceeded the expected values as SDI increased. The fitted curve in the analysis suggests that the expected rate of increase in DALYs burden begins to plateau after an SDI of approximately 0.6. The distribution of countries above and below the expected curve was roughly balanced. Notably, the United States exceeded the expected value and recorded the highest ASDR worldwide in 2021, while Timor-Leste was well below the expected value and had the lowest rate.

Figure 4.

Correlation Analysis between SDI and high BMI-related osteoarthritis. (A) Correlation analysis between SDI and high BMI-related osteoarthritis in 21 GBD regions. (B) Correlation analysis between SDI and high BMI-related osteoarthritis in 204 countries.

ARIMA projections

Using ARIMA modeling, the global ASDR for males and females were projected through 2050 (Fig. 5). Red lines represent the true trend of age-standardized rate of DALYs of high BMI-related osteoarthritis during 1990–2021; yellow dot lines and shaded regions represent the predicted trend and its 95% CI. The results indicate a continued upward trend for both sexes, with the rates estimated to reach 52.395 for males and 73.044 for females per 100 000 population by 2050 (Supplemental Digital Content Table S3, available at: http://links.lww.com/JS9/E833). Throughout the 2021–2050 period, the projected ASDR remains consistently higher in females than in males, further highlighting the disproportionately greater burden of osteoarthritis attributable to high BMI among females.

Figure 5.

The predicted trends of ASDR of osteoarthritis associated with BMI globally over the next 29 years. (A) The predicted trends among males. (B) The predicted trends among females.

MR analysis results

The results of the IVW analysis demonstrated that obesity class 1 (OR = 1.008, 95% CI: 1.005–1.010, P = 4.443 × 10⁻⁸), class 2 (OR = 1.007, 95% CI: 1.005–1.009, P = 7.22 × 10⁻¹¹), and class 3 (OR = 1.007, 95% CI: 1.005–1.009, P = 2.157 × 10⁻⁵) were each causally and positively associated with osteoarthritis. As illustrated in Figure 6, both the IVW and MR-Egger estimates lie to the right of the null line, providing further evidence for a positive causal association. This conclusion is also supported by the findings presented in Supplemental Digital Content Figure S4, available at: http://links.lww.com/JS9/E827. Furthermore, in order to examine the effect of obesity on various forms of osteoarthritis, an analysis was conducted of obesity at differing levels in relation to knee and hip osteoarthritis, first carpometacarpal joint osteoarthritis, and other forms of osteoarthritis. The results indicated that obesity at all levels was associated with the onset and progression of knee and hip osteoarthritis. However, the study did not find any evidence that obesity influenced first carpometacarpal joint osteoarthritis or other types of osteoarthritis. Detailed results for all three obesity classes and osteoarthritis are presented in Supplemental Digital Content Table S4, available at: http://links.lww.com/JS9/E834. Additionally, we performed inverse Mendelian randomization analyses to investigate whether osteoarthritis causally influences obesity (Supplemental Digital Content Table S5, available at: http://links.lww.com/JS9/E835; Supplemental Digital Content Figure S6, available at: http://links.lww.com/JS9/E829; and Supplemental Digital Content Figure S7, available at: http://links.lww.com/JS9/E830). These findings support a unidirectional causal relationship in which varying degrees of obesity contribute to an increased risk of osteoarthritis, while osteoarthritis does not causally influence the development of obesity.

Figure 6.

Forest plot of forward Mendelian randomization analysis. (A) Forward Mendelian randomization analysis of obesity class 1 on osteoarthritis. (B) Forward Mendelian randomization analysis of obesity class 2 on osteoarthritis. (C) Forward Mendelian randomization analysis of obesity class 3 on osteoarthritis.

Sensitivity analyses for heterogeneity in both forward and inverse MR analyses yielded P > 0.05, indicating no evidence of heterogeneity across the studies. Furthermore, the MR-PRESSO global test and MR-Egger intercept tests also produced P > 0.05 in both directions, suggesting the absence of horizontal pleiotropy. These findings support the robustness and reliability of the MR results (Supplemental Digital Content Table S6, available at: http://links.lww.com/JS9/E836; Supplemental Digital Content Table S7, available at: http://links.lww.com/JS9/E837; Supplemental Digital Content Figure S5, available at: http://links.lww.com/JS9/E828; and Supplemental Digital Content Figure S6, available at: http://links.lww.com/JS9/E829).

Discussion

Compared to other studies of high BMI-related osteoarthritis, this study conducted a comprehensive analysis of the disease burden of high BMI-related osteoarthritis, examining trends at global, regional, and national levels from 1990 to 2021 based on GBD 2021 data, with projections for future burden. In addition, the causal relationship between different obesity classes and osteoarthritis was further explored using MR analysis, providing scientific evidence to support future strategies for the prevention and control of osteoarthritis. To our knowledge, this is the first study to explore obesity and osteoarthritis using a combination of GBD 2021 and Mendelian randomization analysis.

Globally, the number of disability-adjusted life years (DALYs) due to high BMI-related osteoarthritis nearly tripled between 1990 and 2021, indicating a substantial rise in disease burden and highlighting the urgent need for more effective disease management and prevention. The extant literature does not contradict this phenomenon, and the increase in the burden of disease may be closely linked to the global increase in obesity prevalence, population aging, and changes in lifestyle patterns^[6,16,17].At the regional level, most studies focus on High-income North America and South Asia because these regions have the highest ASDRs and percentage increase in ASDR, respectively, in the world^[6,16,18]. Interestingly, East Asia consistently reported the highest number of DALYs among all GBD subregions. Although this has been reflected in prior data^[6,16,19,20], it has not been widely discussed. Moreover, the burden of high BMI-related osteoarthritis increased with higher SDI levels, which is consistent with findings from other studies^[6,16,19,21]. High SDI regions carry a heavier disease burden compared to other SDI regions, potentially due to the increase in obesity rates as the standard of living improves with the development of society and technological advances that have increased the detection rate of osteoarthritis^[17,22]. Although the burden is greater in high-SDI regions, the rate of increase is slower. Previous research has suggested an inverse relationship between osteoarthritis burden and factors such as educational attainment and socioeconomic development, which may explain the slower growth in ASDR in these areas^[23–27]. This also raises the possibility of improving osteoarthritis outcomes through public health initiatives aimed at enhancing disease literacy. Age–sex difference analyses showed that the disease burden was consistently higher in females than in males, which is consistent with the known epidemiological characteristics of osteoarthritis and with existing findings^[4,28,29]. This sex disparity may be attributed to differences in hormonal profiles and anatomical structure between males and females^[30,31]. Therefore, the female population should be the focus of more attention in the control of the disease burden of osteoarthritis. Existing research on high BMI-related osteoarthritis has not fully explored the sources of disease burden or strategies for its control^[6]. Our study found that the burden of high BMI-related osteoarthritis is largely driven by population growth and disease prevalence, consistent with the results of existing studies addressing the burden of osteoarthritis, suggesting a strong association between lifestyle factors and the development of osteoarthritis^[19,20]. This connection has been elaborated in prior studies^[32,33]. In terms of disease control, there remains considerable room for improvement in many countries, where the growing challenge posed by obesity and osteoarthritis continues to escalate.

Previous studies have confirmed a causal relationship between BMI and osteoarthritis through MR analyses^[7,8,34,35]. Our research builds upon this foundation by examining the causal effects of different obesity classes on osteoarthritis, thereby providing new insights for prevention and treatment strategies. The causal relationship between various grades of obesity and osteoarthritis was established through forward MR analyses, especially closely related to knee and hip osteoarthritis. Research indicates that obesity significantly increases the risk of developing osteoarthritis and exacerbates clinical symptoms. The likelihood of osteoarthritis is markedly higher in obese individuals compared to those of normal weight. Existing clinical studies reveal that obese osteoarthritis patients exhibit substantial differences in inflammatory status and clinical symptoms, with symptom severity and gait abnormalities being notably more pronounced in obese patients than in those with lower BMI^[36,37]. An animal model study demonstrated that obese mice are more susceptible to osteoarthritis than their normal-weight counterparts, with elevated serum levels of inflammatory factors further corroborating this finding^[38]. Pathologically, obesity primarily influences the development of osteoarthritis through biomechanical and inflammatory mechanisms, specifically by increasing the weight burden on weight-bearing joints and enhancing the expression of inflammatory factors such as IL-1β, TNF-α, COX-2, and MMP-3^[39,40]. This may provide a rationale for the phenomenon observed in this study, wherein the knee and hip joints, as weight-bearing joints of the body, demonstrate a close correlation with obesity.

In terms of treatment options for osteoarthritis, weight management occupies an indispensable position in the current guidelines. Weight loss has been shown to not only reduce the risk of osteoarthritis but also alleviate some symptoms of osteoarthritis^[41–44]. Furthermore, it offers a cost-effective approach in the treatment process, reducing the cost of treating osteoarthritis^[40]. Weight loss has been demonstrated to effectively inhibit the progression of osteoarthritis, including the reduced pain, improved joint function, reduced stiffness, and lowered IL-6. The aforementioned effects are more pronounced when weight loss exceeds 10%^[45–47]. Both non-surgical and surgical weight loss treatments are effective in treating osteoarthritis and are both recommended for obese patients with the condition^[45,48,49]. However, given the invasive nature of surgery, non-surgical treatment is generally preferred for patients in the early stages of obesity, while surgical treatment is typically recommended for those with grade II or III obesity.

Despite the strengths of this study, several limitations should be acknowledged. First, although the data were sourced from the GBD database and underwent cleaning and standardization, potential data inaccuracies remain. There exists a certain discrepancy between GBD data and actual data, particularly in low-SDI regions where diagnostic limitations may lead to underestimated DALY values. Second, this study focused solely on DALYs and ASDR, without assessing incidence or prevalence, leaving a gap in understanding the full epidemiological profile. Third, the number of SNPs identified through MR analysis varied substantially among the three groups, with CLASS 3(11 SNPs) exhibiting significantly fewer SNPs compared to the other two groups (45 and 35 SNPs), potentially indicating reduced statistical power. Finally, other relevant risk factors for osteoarthritis including genetic background, occupational exposure, and environmental influences were not accounted for in this study and warrant further investigation.

Ethical approval

The data used in this study for the burden of disease analysis and Mendelian randomization analyses were obtained from publicly available databases. Hence, the ethical approval was not required.

Consent

All of the data were taken from published sources, and the informed consent and approval were received.

Sources of funding

The study was funded by the National Natural Science Foundation of China (82360420), Science and Technology Fund of Guizhou Science and Technology Department (QKH-ZK[2023]344), Science and Technology Fund of Guizhou Provincial Health Commission (gzwkj2021-261), the Youth Fund cultivation program of National Natural Science Foundation of Affiliated Hospital of Guizhou Medical University (gyfynsfc-2021-12), Graduate Scientific Research Fund project of Guizhou (YJSKYJJ [2021]157), and Doctor Start-up Fund of Affiliated Hospital of Guizhou Medical University (gyfybskj2023-07).

Author contributions

This study was conceived by H.S. and G.X.P. H.S. and L.Z.S. conducted all bioinformatic analyses. M.L. and Y.Z. acquired the data from online database and participated in figures and charts drawing. X.N. and H.Y. supervised the implement of the current study. The funding was provided by H.S., H.Y., and G.X.P. L.Z.S. and M.L. prepared the original draft. H.S., and G.X.P. reviewed and edited the manuscript. The final version of the manuscript was approved by all the authors.

Conflicts of interest disclosure

All authors declare that there is no competing interest.

Guarantor

Hong Sun and Guoxuan Peng.

Research registration unique identifying number (UIN)

Not applicable.

Provenance and peer review

Not commissioned, externally peer-reviewed.

Data availability statement

The original data presented in this study are included in the paper. Moreover, the source data and methodology are listed in supplemental files.