Osteoarthritis burden and inequality from 1990 to 2021: a systematic analysis for the global burden of disease Study 2021

Abstract

Osteoarthritis (OA) is a major global health burden, affecting millions and causing significant disability. Understanding its trends and determinants is crucial for effective management and prevention. We analyzed data from the Global Burden of Diseases (GBD) study 2021 to assess OA incidence, Years Lived with Disability (YLDs), and age-standardized rates (ASIR/ASYR) from 1990 to 2021. We explored trends and determinants across gender, region, and Socio-Demographic Index (SDI) quintiles using Joinpoint regression, Age-Period-Cohort (APC) modeling, decomposition, and inequality analyses. The global incidence of OA surged from 20.9 million in 1990 to 46.6 million cases in 2021, with an AAPC of 0.29%. Correspondingly, YLDs escalated from 8.92 million to 21.30 million, reflecting an AAPC of 0.30%. Disparities exist across SDI quintiles, with higher rates observed in high SDI countries. Women consistently experience a higher burden compared to men. Asian regions demonstrate the fastest rise in ASYR. High BMI contributes significantly to OA burden, particularly in high SDI countries. The rising burden of OA necessitates urgent attention. Interventions targeting modifiable risk factors, such as obesity, and early detection and management strategies are crucial. Addressing gender disparities and health inequalities, particularly in high SDI countries, is essential for effective OA prevention and control.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-025-93124-z.

Introduction

Osteoarthritis (OA) is a leading global cause of disability, affecting 7.96% of the global population in 2020, with a disproportionate impact on middle-aged and elderly individuals¹. Characterized by joint pain, stiffness, and functional impairment, OA significantly diminishes patients’ quality of life^2,3. The global prevalence of OA increased by 132.2% from 1990 to 2020, accompanied by a 9.5% rise in years lived with disability (YLDs)¹, placing a substantial burden on healthcare systems and economies^4,5. OA is a multifactorial disease influenced by age, gender, obesity, genetics, joint injuries, diet, and lifestyle⁶. Obesity, measured by body mass index (BMI), is a significant modifiable risk factor, with the proportion of OA burden attributed to high BMI increasing from 16.1% in 1990 to 20.4% in 2020^7,8. Current management strategies, including pharmacological treatments, physical therapy, and surgery, primarily focus on symptom relief rather than a cure^9,10, and their efficacy varies among patients, with potential adverse effects from prolonged use¹¹. Given these limitations, there is an urgent need for more effective diagnostic, therapeutic, and preventive approaches. This study aims to analyze the global burden and trends of OA to inform enhanced prevention and treatment strategies.

Although previous studies have yielded valuable insights into global trends of OA, they have predominantly concentrated on specific regions¹², time periods⁴, or particular subtypes of the OA¹³. Despite their informative nature, these studies have not offered a comprehensive perspective on the global burden of OA across diverse regions and time frames. Furthermore, they have not thoroughly examined the contributions of various risk factors, such as elevated BMI, to the overall disease burden¹.

Our study offers a comprehensive analysis of the global OA burden using the latest Global Burden of Disease (GBD) 2021 data (1990–2021), aims to address three critical gaps. First, we apply advanced inequality metrics to assess health disparities across Socio-Demographic Index (SDI) levels, an area previously underexplored. Second, our decomposition framework separates the effects of population aging, growth, and epidemiological changes on OA burden, aiding targeted policy actions. Third, we use gender-stratified analyses with age-period-cohort (APC) modeling and joinpoint regression to uncover sex-specific trends over three decades, enhancing previous reports’ detail. This approach advances epidemiological surveillance by identifying modifiable OA burden drivers and informing region-specific prevention strategies.

Methods

Data source

The GBD 2021 employed the latest available epidemiological data and enhanced standardized methods to evaluate health loss comprehensively for 371 diseases, injuries, and impairments, stratified by age and sex across 204 countries and territories¹⁴. This study integrated various data sources, each of which was assigned a unique identifier and included in the Global Health Data Exchange (GHDx)¹⁵. For population data, GBD 2021 used multiple sources, such as national health surveys, censuses, and vital registration systems. Age- and sex-specific prevalence estimates were modeled with DisMod-MR 2.1, which adjusted for biases to standardize the data across different demographics and regions¹⁶. OA and its subtypes (knee, hip, hand, and other joints) were identified using ICD-10 codes¹⁴. The methodologies utilized in GBD 2021 are described earlier¹⁷.

The current investigation assembled data on the incidence, YLDs, and age-standardized incidence rates (ASIR) and YLD rates (ASYR) for OA stratified by sex, region, and country from 1990 to 2021. It also computed YLDs attributable to high BMI. The numerical estimates were presented with 95% uncertainty intervals (95% UI), and the rates were expressed per 100,000 individuals, accompanied by their respective 95% UIs. Notably, as OA-related mortality is not accounted for in the GBD study, this analysis concentrates solely on non-fatal burden metrics, specifically YLDs. Sociodemographic development was measured using the SDI, a composite average of income, educational attainment, and fertility rates¹⁸. The 204 countries and regions were subsequently categorized into five quintiles based on their SDI¹⁹, encompassing low, low-middle, middle, high-middle, and high socio-demographic categories.

Joinpoint analysis

The joinpoint regression model²⁰, a suite of linear statistical techniques, was utilized to assess the time-related patterns in the burden of OA. Utilizing Joinpoint software (Version 4.9.1.0), we estimated age-standardized incidence and YLDs trends for OA, calculating average annual percent change (AAPC) and annual percent change (APC) with their 95% confidential intervals (95%CI), respectively. This approach leverages the least squares method to objectively identify trend shifts and turning points, reducing subjectivity compared to traditional linear trend analysis. The AAPC, computed over the period from 1990 to 2021, quantifies the overall annual rate of change, while APCs assess period-specific trends. An AAPC of 1.0, for instance, indicates a 1.0% annual increase in the OA burden rate. While a 95% CI contains zero, presenting a stable trend over the period.

Age-period-cohort analysis

Our study harnessed the APC model²¹ to meticulously analyze the epidemiological trends of OA among individuals aged 30 to 95 + from 1990 to 2021. The APC model is particularly adept at addressing the issue of collinearity between age, period, and cohort effects, which is a common challenge in epidemiological research. By utilizing the National Cancer Institute’s web tool²², we effectively disentangled these intertwined influences through the estimation of key parameters: Local Drift, capturing the annual change in disease risk within specific age groups; Longitudinal Age Curve, accounting for period deviations to isolate age-specific trends; Period Relative Risks, comparing age-specific risks across periods to the reference year of 2002; and Cohort Relative Risks, contrasting risks between cohorts relative to the 1950 reference cohort. The model’s sophisticated approach for assessing the significance of trends, provided a robust framework for understanding the complex determinants of OA incidence and YLDs, thereby enhancing our ability to measure disease burdens, identify risk factor patterns, and inform public health strategies.

Decomposition analysis

In this study, we conducted a decomposition analysis to elucidate the contributions of aging, population growth, and epidemiological changes to the YLDs and incidence rates of OA across various anatomical sites. Utilizing the Das Gupta methodology²³, we dissected the impact of age structure, population dynamics, and disease-specific rates on YLDs from 1990 to 2021. Further decomposition was performed for regions with significant epidemiological shifts, isolating the effects of subtypes of OA including knee, hip, hand and others within age groups. The resulting proportions of change attributable to each factor provided insights into the underlying drivers of the observed trends.

OA burden attributed to high BMI

In GBD 2021, high BMI (≥ 25 kg/m²) was the only risk factor for OA examined. Briefly, population exposure to high BMI was estimated for each nation, age, gender, and year through a hybrid approach involving spatiotemporal Gaussian process regression and mixed-effects models. The burden attributed to high BMI was determined by multiplying the overall YLDs of OA by the population attributable fraction (PAF)²⁴. The PAF measures the percentage of YLDs that might be diminished if the population’s exposure to the risk factor was minimized to the lowest hypothetical risk level.

Cross-country inequalities analysis

To assess the socio-economic disparities in the burden of OA across countries, the Slope Index of Inequality (SII) and Concentration Index (CI)²⁵ were applied as analytical tools. The SII was derived from a regression analysis of country-level OA incidence and YLDs across all age groups on a relative position scale, reflecting sociodemographic development and defined by the midpoint of the cumulative class range based on SDI rankings. The CI was computed by adapting a Lorenz curve to the cumulative proportional distribution of populations according to SDI, incorporating disease incidence and YLDs, and numerically integrating the area beneath the curve²⁶.

All analytical and graphical presentation was conducted by R software, version 4.3.3, incorporating the MASS and ggplot2 packages. The statistical models provided results with 95% confidence intervals (CIs), and we considered a p-value threshold of less than 0.05 to denote statistical significance.

Results

Overall trend of OA globally

Globally, the incidence of the disease increased from 20,900,510 cases in 1990 (95% UI: 18,467,652.64 to 23,104,315.92) to 46,632,144 cases in 2021 (95% UI: 41,122,052.73 to 51,644,430.77) (Table 1, S1). Concurrently, the ASIR per 100,000 population also rose from 489.78 in 1990 (95% UI: 433.10 to 541.51) to 535.00 in 2021 (95% UI: 472.38 to 591.97). Similarly, the global YLDs increased from 8,918,856.67 years in 1990 (95% UI: 4,264,150.65 to 17,983,775.68) to 21,304,565.82 years in 2021 (95% UI: 10,189,161.47 to 42,935,419.74) (Table 2, S2). The ASYR per 100,000 population also presented an uptrend, from 222.80 in 1990 (95% UI: 106.65 to 450.29) to 244.50 in 2021 (95% UI: 117.06 to 493.11).

Table 1.

Incidence for osteoarthritis: age-standardized rates with 95% uncertainty intervals and average annual percentage change (AAPC) with 95% confidence intervals, 1990–2021.

	1990		2021		1990–2021
	Incidence cases No. *105 (95% UI)	ASIR per 100,000 No. (95% UI)	Incidence cases No. *105 (95% UI)	ASIR per 100,000 No.(95% UI)	AAPC No. (95% CI)
Global	209.01[184.68 to 231.04]	489.78 [433.10 to 541.51]	466.32 [411.22 to 516.44]	535.00 [472.38 to 591.97]	0.29 [0.28 to 0.30]
Female	123.97[109.71 to 136.67]	568.60 [502.58 to 627.86]	277.03 [244.80 to 305.67]	621.32 [548.49 to 686.94]	0.28 [0.25 to 0.31]
Male	85.04[75.01 to 94.51]	408.56 [361.23 to 452.94]	189.29 [166.88 to 210.24]	445.74 [393.68 to 493.98]	0.28 [0.27 to 0.29]
High SDI	58.65[52.26 to 64.79]	568.82 [505.90 to 627.74]	101.89 [90.89 to 112.69]	611.30 [542.71 to 675.91]	0.23 [0.19 to 0.27]
Female	35.61[31.72 to 39.18]	655.38 [582.19 to 722.51]	60.17 [53.71 to 66.28]	710.08 [630.20 to 781.80]	0.26 [0.21 to 0.30]
Male	23.03[20.44 to 25.59]	475.17 [421.30 to 527.48]	41.72 [37.10 to 46.57]	511.67 [455.01 to 568.79]	0.24 [0.21 to 0.27]
High middle SDI	51.34[45.38 to 56.75]	498.45 [441.11 to 549.27]	103.50 [90.99 to 114.68]	548.07 [481.66 to 608.49]	0.31 [0.29 to 0.32]
Female	31.20[27.65 to 34.43]	574.26 [507.89 to 633.60]	62.23 [54.80 to 68.75]	638.30 [559.69 to 707.96]	0.34 [0.33 to 0.36]
Male	20.14[17.75 to 22.35]	415.59 [368.00 to 461.36]	41.27 [36.28 to 46.13]	453.75 [399.69 to 505.25]	0.28 [0.26 to 0.30]
Middle SDI	57.51[50.58 to 64.10]	475.42 [419.58 to 527.87]	154.38 [135.69 to 171.43]	536.49 [473.16 to 595.02]	0.40 [0.38 to 0.42]
Female	33.46[29.45 to 37.28]	557.05 [491.55 to 617.57]	91.98 [80.91 to 101.87]	626.35 [552.23 to 692.98]	0.38 [0.36 to 0.40]
Male	24.05[21.16 to 26.83]	395.11 [348.79 to 440.06]	62.40 [54.92 to 69.75]	443.48 [391.42 to 492.86]	0.37 [0.35 to 0.39]
Low middle SDI	30.35[26.75 to 33.76]	423.51 [374.68 to 469.50]	78.12 [68.92 to 86.77]	480.13 [424.96 to 533.01]	0.41 [0.40 to 0.42]
Female	17.44[15.38 to 19.35]	494.78 [437.00 to 548.78]	46.39 [40.86 to 51.55]	557.53 [491.70 to 617.08]	0.38 [0.37 to 0.40]
Male	12.91[11.33 to 14.43]	355.29 [315.57 to 394.88]	31.73 [28.01 to 35.41]	399.88 [354.76 to 443.93]	0.39 [0.37 to 0.40]
Low SDI	10.95[9.66 to 12.19]	408.88 [362.19 to 452.55]	28.05 [24.79 to 31.23]	447.12 [395.36 to 493.37]	0.29 [0.29 to 0.30]
Female	6.13[5.41 to 6.83]	464.16 [410.04 to 513.34]	16.03 [14.16 to 17.82]	503.73 [443.82 to 559.60]	0.27 [0.26 to 0.28]
Male	4.81[4.23 to 5.38]	355.37 [314.53 to 396.66]	12.02 [10.62 to 13.43]	389.25 [345.36 to 432.19]	0.30 [0.29 to 0.31]

ASIR, age standardized incidence rate; AAPC, average annual percent change; SDI, Socio-Demographic Index; 95% UI, uncertainty interval; 95% CI, 95% confidential interval.

Table 2.

Years lived with disability (YLDs) for osteoarthritis: age-standardized rates with 95% uncertainty intervals and average annual percentage change (AAPC) with 95% confidence intervals, 1990–2021.

	1990		2021		1990–2021
	YLDs (Years Lived with Disability) No. *105 (95% UI)	ASYR per 100,000 No. (95% UI)	YLDs (Years Lived with Disability) No. *105 (95% UI)	ASYR per 100,000 No.(95% UI)	AAPC No. (95% CI)
Global	89.19[42.64 to 179.84]	222.80 [106.65 to 450.29]	213.05 [101.89 to 429.35]	244.50 [117.06 to 493.11]	0.30 [0.26 to 0.33]
Female	54.85[26.23 to 110.75]	257.65 [123.39 to 520.40]	130.21 [62.41 to 262.72]	284.14 [136.29 to 573.11]	0.31 [0.27 to 0.35]
Male	34.34[16.42 to 69.08]	182.42 [87.58 to 369.04]	82.83 [39.57 to 166.71]	200.52 [95.88 to 404.87]	0.31 [0.29 to 0.32]
High SDI	28.45[13.66 to 57.25]	262.74 [126.00 to 529.15]	54.87 [26.49 to 110.52]	283.13 [136.04 to 570.53]	0.23 [0.19 to 0.28]
Female	18.14[8.72 to 36.46]	298.96 [143.22 to 602.03]	33.87 [16.38 to 68.48]	328.75 [157.93 to 662.63]	0.31 [0.27 to 0.34]
Male	10.32[4.94 to 20.80]	217.21 [104.21 to 437.89]	21.00 [10.10 to 42.31]	232.74 [111.74 to 469.20]	0.23 [0.20 to 0.26]
High middle SDI	22.95[10.92 to 46.22]	228.99 [109.20 to 462.13]	49.63 [23.67 to 99.75]	250.58 [119.78 to 503.69]	0.29 [0.28 to 0.31]
Female	14.62[6.95 to 29.51]	262.19 [124.59 to 529.06]	30.88 [14.76 to 62.34]	290.53 [138.73 to 585.88]	0.33 [0.31 to 0.36]
Male	8.33[3.98 to 16.80]	186.55 [89.46 to 376.52]	18.75 [8.94 to 37.47]	204.71 [97.59 to 410.36]	0.30 [0.29 to 0.31]
Middle SDI	22.35[10.77 to 44.86]	208.27 [100.47 to 419.68]	67.08 [32.04 to 134.66]	240.41 [115.09 to 483.99]	0.47 [0.46 to 0.48]
Female	13.21[6.35 to 26.48]	241.76 [116.44 to 486.54]	40.65 [19.38 to 81.71]	279.78 [133.59 to 563.51]	0.47 [0.45 to 0.50]
Male	9.15[4.41 to 18.38]	173.17 [83.27 to 349.11]	26.43 [12.66 to 53.19]	197.58 [94.69 to 398.39]	0.42 [0.39 to 0.45]
Low middle SDI	11.36[5.47 to 22.79]	180.18 [86.93 to 364.19]	31.11 [14.92 to 62.49]	209.35 [100.40 to 422.62]	0.49 [0.48 to 0.50]
Female	6.59[3.16 to 13.21]	212.26 [102.05 to 428.24]	18.83 [9.03 to 37.70]	244.67 [117.48 to 491.31]	0.46 [0.45 to 0.47]
Male	4.77[2.31 to 9.59]	148.95 [72.21 to 301.48]	12.27 [5.90 to 24.83]	171.16 [81.93 to 348.15]	0.45 [0.44 to 0.47]
Low SDI	3.98[1.93 to 7.99]	170.90 [82.56 to 345.11]	10.18 [4.89 to 20.37]	190.93 [91.63 to 384.26]	0.36 [0.35 to 0.37]
Female	2.24[1.08 to 4.50]	195.44 [93.68 to 395.34]	5.87 [2.81 to 11.72]	216.58 [103.39 to 434.29]	0.33 [0.31 to 0.36]
Male	1.75[0.85 to 3.52]	146.99 [71.69 to 296.87]	4.31 [2.08 to 8.67]	164.09 [79.31 to 330.89]	0.36 [0.34 to 0.37]

YLDs, Years Lived with Disability ; ASYR, age standardized Years Lived with Disability rate; AAPC, average annual percent change; SDI, Socio-Demographic Index; 95% UI, uncertainty interval; 95% CI, 95% confidential interval.

The Joinpoint model indicated that both the ASIR (AAPC = 0.29, 95% CI: 0.28 to 0.30) and ASYR (AAPC = 0.3, 95% CI: 0.26 to 0.33) exhibited upward trends globally (Figs. 1 and 2), with a greater rate of change observed after 2000. The ASIR and ASYR in high SDI countries remained the highest among the quintiles over the 31-year period, reaching 535.00 (95% UI: 472.38 to 597.97) and 283.13 (95% UI: 136.04 to 570.53) in 2021, respectively.

Fig. 1.

Global trends of ASIR of OA stratified by SDI quintiles from 1990 to 2021. OA, osteoarthritis; ASIR, age standardized incidence rate; AAPC, average annual percent change; APC, annual percent change; SDI, Socio-Demographic Index; 95% CI, 95% confidential interval; *, significant change.

Fig. 2.

Global trends of ASYR of OA stratified by SDI quintiles from 1990 to 2021. OA, osteoarthritis; ASYR, age standardized Years Lived with Disability rate; AAPC, average annual percent change; APC, annual percent change; SDI, Socio-Demographic Index; 95% CI, 95% confidential interval; *, significant change.

Regarding the ASIR, the upward trends in the high-middle and middle SDI quintiles accelerated significantly between 2000 and 2005 (APC = 1.01, 95% CI: 0.96 to 1.06 and APC = 1.02, 95% CI: 0.97 to 1.08, respectively), while the high SDI quintile experienced a slight decline from 1995 to 2005 (APC= -0.16, 95% CI: -0.18 to -0.13) before resuming its upward trajectory. The ASYR presented a similar pattern of change across SDI quintiles. Notably, the low-middle SDI quintile showed the largest increase in both ASIR (AAPC = 0.29, 95% CI: 0.29 to 0.30) and ASYR (AAPC = 0.49, 95% CI: 0.48 to 0.50) over the 31-year period.

When stratified by gender, women consistently had higher ASIR and ASYR than men across all regions, with both rates showing an upward trend (Figure S1, S2). Globally, the ASIR for women remained stable in recent years (2014–2021 APC = 0.03 [95% CI contained zero]), whereas men experienced a continuous rise over 31 years (AAPC = 0.28, 95% CI: 0.27 to 0.29). Analyzing the SDI quintiles, only the high SDI quintile showed no recent increase for women, with a slight decline from 2015 to 2018 (APC = − 0.54, 95% CI: − 0.87 to − 0.21) followed by stability from 2018 to 2021 (APC = 0.17, 95% CI: − 0.002 to 0.35). The ASYR exhibited a similar pattern of change. Specifically, for the high SDI quintile, women experienced a significant decrease from 2015 to 2019 (APC = − 1.03, 95% CI: − 1.19 to − 0.87).

Among the 21 regions, the fastest rise in ASYR was observed in Southeast Asia (AAPC = 0.59, 95% CI: 0.58 to 0.61), South Asia (AAPC = 0.57, 95% CI: 0.58 to 0.61), and North Africa and Middle East (AAPC = 0.53, 95% CI: 0.50 to 0.57). Similarly, the fastest increase in ASIR was found in Southeast Asia (AAPC = 0.49, 95% CI = 0.48 to 0.50), North Africa and Middle East (AAPC = 0.46, 95% CI = 0.44 to 0.48), and South Asia (AAPC = 0.45, 95% CI = 0.44 to 0.46). Notably, the Asian regions consistently ranked among the top for the rate of increase in both ASYR and ASIR.

In the analysis of 204 countries and territories, the Republic of Korea, Singapore, and Brunei Darussalam consistently ranked among the top three for both ASIR and ASYR in 1990 and 2021 (Figure S3). The Republic of Korea held the first position for ASIR in both 1990 and 2021, while its ASYR ranking rose from second in 1990 to first in 2021. The countries with the largest increases in ASIR were Equatorial Guinea, Mongolia, and Thailand, whereas the largest increases in ASYR were observed in Equatorial Guinea, Mongolia, and Ethiopia (Figure S4). Notably, Denmark was the only country globally where both ASIR (AAPC= − 0.02, 95% CI: − 0.05 to − 0.002) and ASYR (AAPC = − 0.12, 95% CI: − 0.13 to − 0.10) showed an overall downward trend over the 31-year period.

The four OA subtypes—hand, hip, knee, and other OA—represented distinct proportions of the total burden. In 2021, knee OA accounted for the largest proportion of both incidence cases (66.14%) and YLDs (56.41%), followed by hand OA, which represented 22.23% of incidence and 28.95% of YLDs. Other OA contributed to 7.77% of incidence and 9.29% of YLDs, while hip OA had the smallest proportion, with 3.85% of incidence and 5.35% of YLDs.

Significant regional disparities in ASIR (per 100,000) of each OA subtypes were observed across 21 regions in 2021. High-income Asia Pacific had the highest ASIR for knee OA at 458.22, significantly above the global level. Eastern Europe exhibited the highest ASIR for hand OA at 209.49, surpassing the global level as well. High-income North America reported a notably high rate of 45.62 for hip OA. For other OA, Southern Sub-Saharan Africa had the highest ASIR at 44.85. Detailed disease burden and AAPC values for all 204 countries and territories are provided in Table S3 to S10 and Figure S5 to S12.

APC model

The incidence of OA showed a positive local drift, indicating a steady increase both globally and across all five SDI quintiles (Fig. 3A). The annual change in incidence followed an age-dependent S-shaped curve, with the highest point mostly occurring in the 35–49 age group. Significant differences in annual change were observed among age groups, but the low SDI curve was relatively smooth, indicating smaller differences between ages. In the high SDI quintile, the annual change for males over 50 years was consistently greater than for females, unlike in other regions where females had higher rates.

Fig. 3.

Estimates of age, period, and cohort effects on incidence of osteoarthritis globally and in different SDI regions. (A) Local drift, the horizontal colored lines represent the net drift of different groups; (B) Age effect; (C) Period effect; (D) Cohort effect.

In terms of age effect, the raw incidence rate of OA followed an inverted U-shape, with females always having higher rates than males (Fig. 3B). The highest incidence rate was mainly seen in the 50 to 64 age group, where the difference between genders increased, and this difference grew with higher SDI.

The period effect analysis showed an upward trend in the incidence rate ratio for the global population and across all SDI quintiles (Fig. 3C). After 2002, the gender difference first increased and then decreased in recent years. The birth cohort effect showed a global upward trend in the risk ratio of OA incidence, with females consistently having higher rates than males (Fig. 3D).

The patterns of local drift, period effect, and cohort effect on YLDs were similar to those of incidence, except for middle SDI females, whose annual change in YLDs remained high without a clear peak (Figure S13). The age effect on YLDs showed a continuous increase with age, with the gender difference growing progressively and associated with higher SDI levels.

Decomposition analysis

To investigate the impact of population growth, aging, and epidemiologic changes including four OA subtypes (hip, knee, hand, and others) on the epidemiological landscape of OA over a 31-year span, we conducted a decomposition analysis of crude incidence and YLDs across different populations, age structure, and standardized rates for age and population (referred to herein as epidemiological transitions). Across the board, there was a marked escalation in the incidence of OA on a worldwide scale and within each SDI quintile, with the most substantial surge observed in the middle SDI quintiles, where the aggregate incidence rise was most evident (Fig. 4A). Globally, aging of the world population followed by population growth contributed 9.13% and 79.50%, respectively, to the increased burden of incidence between 1990 and 2021 (Table S11). The contribution of population growth to overall incidence was most pronounced in the low-SDI quintile (92.30%). The contribution of aging was the highest in high-SDI quintile (15.31%), yet decreased in low-middle SDI (− 2.69%). Epidemiology change of knee and hand contributed most of the epidemiology change in OA incidence in the four subcategories. In high middle SDI quintile, epidemiology change of knee contributed the most with 9.87%, while in middle SDI quintile, epidemiology change of hand contributed the most (7.43%). The contribution pattern of YLDs also presented the same with incidence (Fig. 4B, Table S12).

Fig. 4.

Decomposition of changes in incidence (A) and YLDs (B) of OA globally and by SDI Quintile. A positive magnitude indicates an increase in numbers attributable to the component, a negative magnitude indicates a decrease in attribution, and a black point represents the overall number. OA, osteoarthritis; YLDs, Years Lived with Disability; SDI, Socio-Demographic Index.

OA burden attributed to high BMI

Globally, the proportion of YLDs due to high BMI increased from 16.15% in 1990 to 20.69% in 2021 (Fig. 5A). Across SDI quintiles, the highest population attributable fraction (PAF) for high BMI was observed in the highest SDI quintile, rising from 20.03% in 1990 to 23.56% in 2021 (Fig. 5B). The most substantial increase in PAF was noted in the low-middle SDI quintile, with a 1.54-fold rise, followed by the middle SDI quintile, with a 1.45-fold increase. In 2021, the top three regions with the highest PAF were Australasia, Southern Latin America, and North Africa and the Middle East. With respect to gender, while PAFs increased uniformly across countries, a convergence was observed between the high-middle and high SDI quintiles among females, primarily due to fluctuations in the PAF of the high SDI quintile.

Fig. 5.

Burden and trends of OA attributable to high BMI. (A) Age-standardized YLD rates of OA due to high BMI globally, across 5 SDI levels, and in 21 regions in 1990 and 2021. (B) Changes in age-standardized YLD rates of OA associated with high BMI from 1990 to 2021. BMI, body mass index; OA, osteoarthritis; YLDs, years lived with disability; SDI, Sociodemographic Index.

Cross country inequality analysis

Significant disparities in the burden of OA, both in absolute and relative terms, were associated with variations in SDI, manifesting as a notable rise in the SII over the years. A disproportionate concentration of higher incidence and YLDs was evident in nations with a higher SDI. The SII for incidence in 1990 stood at 436.57, indicating an excess of 436.57 (per 100,000 individuals) incident cases in the most affluent SDI country compared to the least affluent in the same year (Fig. 6A). By 2021, this disparity had widened, reaching a value of 678.05. The SII of YLDs was 229.66 in 1990 and increased to 402.82 in 2021 (Fig. 6B). However, the concentration index of both measures showed slightly change from 1990 to 2021 (Fig. 6C, D).

Fig. 6.

Health inequality regression curves (A, B) and concentration curves (C, D) for the crude incidence and YLDs rate of osteoarthritis worldwide, in 1990 and 2021. SII, slope index of inequality; CI, concentration index; YLDs, Years Lived with Disability; OA, osteoarthritis.

Discussion

This comprehensive analysis of the global burden of OA from 1990 to 2021 revealed a significant rise in incidence and YLDs globally. The ASIR and ASYR rates exhibited an increasing trend across all SDI quintiles, peaking in high SDI nations. A pronounced gender disparity was evident, with women consistently exhibiting higher age-standardized rates than men across all regions. The decomposition analysis indicated that knee and hand OA are the predominant factors behind the epidemiologic shifts, especially in middle and high-middle SDI quintiles. The primary drivers of the escalating OA burden were population growth, aging, and high BMI, with the most substantial inequalities observed in higher SDI countries.

Joinpoint analysis, elucidating the segmented trends in numerical shifts, indicated a substantial global rise in the ASIR and ASYR for OA from 1990 to 2021. Notably, both rates exhibited a steeper escalation post-2000, with a tendency towards a flattened slope from 2015 onwards. This pattern potentially correlates with the aging global demographic, the burgeoning obesity epidemic, and enhanced diagnostic medical technology coupled with increased prevention awareness, especially in the wake of the new millennium^27,28. Our analysis identified fluctuations in the OA burden curve around the year 2000 across various SDI quintiles, potentially resulting from the interplay between OA prevention initiatives and factors such as population aging²⁹ and detrimental lifestyles³⁰.

Women across all SDI quintiles consistently bore a greater OA burden. This discrepancy can be attributed to several factors. Firstly, estrogen’s protective effect on articular cartilage diminishes post-menopause, potentially elevating the risk of OA³¹. Secondly, female joint structure may be inherently more susceptible to OA development compared to males⁹. Additionally, some findings suggested that genetic factors such as KIAA1210 and CaMK4, both with higher expression in female OA patients, may contribute to OA development through pathways involving TGF-β signaling, immune response, and inflammation^32,33. Women’s greater engagement in household labor and caregiving responsibilities may also contribute to a heightened risk of joint injury and OA³⁴. Furthermore, delayed medical consultation by women could result in postponed diagnosis and treatment of OA, exacerbating the YLDs burden³⁵. Notably, the ASYR for women in high SDI countries exhibited a slight decline from 2015 to 2018 before stabilizing, which may reflect increased health awareness, lifestyle improvements, and enhanced healthcare access among women in these regions.

The most rapid increase in ASYR is predominantly observed in Asia, encompassing Southeast Asia and South Asia. This phenomenon may be attributed to multiple factors, with the rising prevalence of obesity in this region being a probable key driver. The marked rise in BMI in South and Southeast Asia, concurrent with economic growth, has led to a corresponding surge in OA burden²⁸.Studies indicate a higher prevalence of knee OA (KOA) among Japanese women, possibly due to frequent squatting, compared to their American counterparts³⁶. Additionally, environmental factors such as higher barometric pressure and relative humidity are positively associated with the intensity of OA pain, suggesting that these conditions may exacerbate pain sensations in individuals with OA³⁷. Ethnic disparities also contribute to the varying OA burden across regions, with Chinese women exhibiting a 45% elevated prevalence of knee OA relative to their white women³⁸. Furthermore, economic globalization, characterized by the shift of factories from developed to less developed countries, has spurred industrial growth and increased labor demand. However, the mismatch between economic advancement and healthcare access may be contributing to the steep increase in OA incidence in Asia and North Africa³⁹. These findings underscore the urgent need for the development of more targeted prevention and management strategies in the Asian region to address the growing burden of OA.

Our study intriguingly revealed a downward trend in both ASIR and ASYR for osteoarthritis in Denmark, in stark contrast to the global upward trend. This discrepancy may be ascribed to the implementation of the Good Life with osteoArthritis from Denmark (GLA: D®) program⁴⁰. Denmark’s GLA: D® program thrives on an evidence-based approach, combining patient education and supervised exercise therapy, which have been proven effective for osteoarthritis⁴⁰. The program’s structured 8-week format, delivered by certified practitioners, ensures consistency and scalability⁴⁰. Its patient-centered design, emphasizing empowerment and self-management, integrates seamlessly into the healthcare system, supported by continuous monitoring and evaluation for ongoing improvement⁴⁰. For low-SDI countries, successful adaptation hinges on training community health workers, ensuring cultural relevance, securing sustainable funding, and strong policy endorsement for non-pharmacological interventions, integrated within existing health frameworks. A significant proportion of patients have experienced clinically meaningful pain reduction, underscoring the program’s potential for global dissemination.

The APC model not only confirmed our earlier findings but also provided further insights. Age effect revealed that the global burden of OA escalated with age, attributed to progressive joint-specific structural degeneration, compromised cellular reparative functions, and defective tissue remodeling. These elements diminish the capacity for cartilage repair, culminating in an irreversible decline in joint integrity, for which a definitive cure remains elusive^9,10. Local drift showed that the global peak in the annual percentage change of OA incidence is predominantly observed among individuals aged 35–49, indicating the fastest rate of uprising trend among all age groups and necessitating immediate attention. The rapid rise in OA incidence among individuals aged 35–49 is primarily driven by the rising prevalence of overweight and obesity^41,42, sports-related joint injuries⁴³, and the increasing adoption of unhealthy lifestyles⁴⁴ among young and middle-aged adults, highlighting the urgent need for preventive measures. Notably, the model indicates the highest annual increase in OA incidence among 35–49 age group, but the overall incidence still peaks in older age groups, following an inverted U-shape pattern. In the high SDI quintile, males over 50 years exhibited a greater annual change in incidence compared to females, contrasting with other regions where females had higher rates. Potentially explained by the fact that in high-SDI regions, with a focus on women’s OA and effective prevention and management strategies, the rate of change in males might be relatively elevated. Additionally, the unhealthy lifestyle habits of men over 50 in these regions, including modern dietary patterns that predispose to obesity and a lack of physical activity⁴⁵, could be another contributing factor. A trend of initially rising and subsequently declining gender differences in OA incidence rates was observed after 2002, potentially associated with heightened gender-specific health awareness, the implementation of preventive strategies, and progress in medical technology. Regarding the birth cohort effect, this study identified a higher burden of OA among later-born individuals compared to their earlier-born counterparts. Despite the greater emphasis on health education and improved medical care, these factors did not offset the rising trend in OA incidence attributable to population growth and aging. Consequently, implementing early detection protocols, gender-specific OA management strategies, and regional surveillance systems should be advocated to monitor the escalating burden of OA over time.

Our decomposition analysis indicated that population growth predominantly influenced the overall incidence rate in the lowest SDI quintile, reflecting rapid demographic expansion in younger populations with limited aging effects, whereas aging had the most significant impact in the highest SDI quintile, driven by prolonged life expectancy and a larger proportion of older adults. Critically, this analysis quantifies the independent contributions of drivers under counterfactual scenarios—population growth and aging are foundational but context-dependent, neither universally dominant nor mutually exclusive. Notably, changes in the epidemiology of KOA and hand OA were the primary contributors to the observed epidemiological shifts in OA incidence across four subcategories, corroborating findings from previous studies^9,46. While hip OA is a significant contributor to the OA burden, it is not the most predominant subtype driving epidemiological changes. Knee OA, with its higher prevalence, greater susceptibility to biomechanical stress, more severe clinical presentation, and stronger association with obesity and joint injuries⁴⁶, accounts for a larger proportion of incidence and YLDs. Additionally, the economic impact of knee OA is substantial⁴⁶, further highlighting its dominant role in shaping OA epidemiology. In the high middle SDI quintile, the epidemiological changes in KOA predominated, potentially due to the higher prevalence of obesity⁴⁷, a recognized risk factor for knee OA⁴⁵. Conversely, in the middle SDI quintile, hand OA contributed most significantly to epidemiological changes, which may be attributed to the more frequent engagement in handcrafting and heavy manual labor, activities that heighten the risk of hand arthritis through repetitive use and hand stress^48,49. These findings highlight that modifiable risks (e.g., obesity, occupational hazards) amplify OA burden independently of demographic trends, particularly in middle- and high-middle SDI regions. Collectively, the results provide a novel perspective for comprehending the epidemiology of OA, suggesting that interventions should be tailored regionally: low-SDI regions require population control strategies, high-SDI regions need geriatric-focused healthcare, and industrializing areas must prioritize workplace ergonomics and weight management. Epidemiological changes, including enhanced diagnostics and increased awareness, have led to higher reported OA incidence rates, reflecting its true prevalence. Concurrent lifestyle shifts, notably rising obesity and sedentary behavior, have expanded the at-risk population, exacerbating the disease burden. Occupational and environmental factors further complicate the issue, underscoring the need for a comprehensive approach to OA management and prevention.

OA due to high BMI consistently accounted for a significant portion of the overall OA burden, with varying growth rates across the five SDI quintiles. Obesity exacerbates OA through mechanical joint overload and systemic inflammation induced by adipose tissue⁹. Our study comprehensively analyzes the global burden of OA due to high BMI, examining trends and regional differences to provide insights for targeted interventions and policy-making. Interventions targeting weight reduction in overweight individuals have been associated with a decreased incidence of knee OA, illustrating the potential of weight management in alleviating the OA burden⁵⁰. Regions with a high PAF for high BMI should prioritize strategies to reduce obesity, such as promoting healthy diets and increasing physical activity. Engaging in joint-friendly physical activities, such as swimming and cycling, has also been linked to improved joint health and reduced OA-related pain, further mitigating the burden⁵¹. Adoption of a healthy diet, incorporating specific strategies like calorie restriction and portion control, is vital for effective weight management in OA patients⁵². Consequently, developing policies that address the distribution and temporal trends of obesity-related OA in different regions is imperative.

Our findings revealed a larger disparity in the SII with a minimal difference in the CI between 1990 and 2021, indicating more pronounced health inequalities, while the imbalanced distribution globally remained unchanged. Regions with higher SDI typically possess superior medical resources and health management, which can mitigate the disease burden. However, this study revealed that high SDI areas carry a disproportionately greater burden of OA, with this disparity widening from 1990 to 2021. High SDI countries, despite their advanced healthcare resources, face a significant burden of OA, potentially due to the high prevalence of obesity and an aging population, both established risk factors for OA⁵³. This burden is exacerbated by a focus on advanced treatments like joint replacement rather than prevention and early diagnosis. Moreover, detrimental lifestyle practices such as contemporary eating habits that increase the risk of obesity and insufficient physical activity also exacerbate the osteoarthritis burden in these areas⁴⁵. In contrast, regions with low SDI exhibited the least OA burden, as indicated by the lowest AAPC value, a phenomenon likely influenced by their population structure. Lower SDI quintiles typically feature younger populations due to higher birth rates and lower life expectancy. The scarcity of medical resources and limited health awareness in these areas also impeded early disease management⁵⁴. Moreover, a lifestyle marked by more physical labor and lower obesity rates²⁷ in low SDI regions may contribute to the slower growth in OA incidence. These insights are pivotal for shaping global health policies and strategies, necessitating recognition of health inequalities and efforts to bridge the gap between nations. In high SDI regions, it is essential to capitalize on their medical resource advantages while focusing on the challenges posed by aging populations and obesity prevalence. Concurrently, support for lower SDI countries should be intensified to elevate healthcare standards, enhance health literacy, and promote lifestyles that alleviate the OA burden. Ultimately, the global promotion of effective prevention and management strategies is crucial to ensure equitable access to advanced medical technologies and practices.

Our study should be interpreted with several limitations in mind. Firstly, case counts in less developed regions within the GBD might be lower than actual figures due to suboptimal healthcare performance, potentially leading to missed diagnoses and documentation gaps. Secondly, reliance on data with inherent heterogeneity and the use of model-generated estimates introduce potential biases and accuracy constraints. Thirdly, the study’s emphasis on BMI as a risk factor, along with limited exploration of OA subtypes and other risk factors, suggests the need for further research. Lastly, the temporal lag in data and the absence of regional and cultural contextualization limit the study’s overall comprehensiveness. It should be noted that, the GBD study employs a sophisticated sampling and weighting methodology to ensure global representativeness. This approach helps to minimize but cannot entirely eliminate potential biases. Further, more detailed investigations are still needed to fully understand the nuances of OA distribution and its determinants, particularly in relation to urban-rural differences and the representativeness of surveys at the national level.

Conclution

In conclusion, this study indicated a substantial increase in both incidence and YLDs, with the highest burden observed in high SDI countries, underscoring the impact of obesity and an aging population in these regions. The gender disparity is evident, with women bearing a heavier burden across all SDI quintiles, likely due to a combination of biological, genetic, and lifestyle factors that contribute to a higher risk and later diagnosis of OA. Furthermore, the study illuminates the significant health inequalities, with a widening gap in the SII from 1990 to 2021, indicating that while higher SDI regions have better medical resources, they paradoxically face a more considerable disease burden. This points to the need for a more equitable distribution of healthcare resources and the implementation of effective prevention and management programs globally. The findings emphasize the urgency of addressing the modifiable risk factors, such as high BMI, and the critical role of timely detection and intervention strategies that are sensitive to the unique challenges faced by different demographic groups. By acknowledging these disparities and health inequalities, the global health community can work towards more inclusive and effective policies that alleviate the burden of osteoarthritis worldwide.

Electronic Supplementary Material

Below is the link to the electronic supplementary material.

Abbreviations

OA

Osteoarthritis

YLDs

Years lived with disability

ASIR

Age-standardized incidence rates

ASYR

Age-standardized years lived with disability rates

GBD

Global burden of disease

SDI

Socio-demographic index

UI

Uncertainty intervals

APC

Annual percent change

AAPC

Average annual percent change

PAF

Population attributable fraction

Author contributions

Ruofeng Wu: Conceptualization; Methodology; Data curation; Visualization; Writing - original draft; Formal analysis. Yu Guo: Conceptualization; Investigation; Methodology; Software; Visualization; Formal analysis.Yi Chen: Methodology; Software; Validation; Data curation. Jingwen Zhang: Methodology; Statistics; Validation; Formal analysis; Writing - review & editing; Supervision.

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

Data availability

The datasets analysed during the current study are available in the GBD 2021 online repository (http://ghdx·health data·org/gbd-results-tool).”

Declarations

Competing interests

The authors declare no competing interests.

Consent of data

Given that the research solely entailed data analysis from the GBD 2021 study and did not involve human subjects, the need for informed consent and institutional review board approval was waived.