Global prevalence, incidence, and years lived with disability (YLDs) of osteoarthritis: trends from 1990 to 2021 and projections to 2050

Abstract

Background

To comprehensively assess the global burden of osteoarthritis (OA) from 1990 to 2021, analyze trends and patterns in its epidemiological shifts, and to project its changes through 2050.

Methods

Data were extracted from the Global Burden of Disease Study 2021, including the numbers and age-standardized rates (ASRs) of prevalence, incidence, and years lived with disability (YLDs) of OA. We described OA epidemiology and trends by sex, age, anatomic site, high body mass index (BMI), sociodemographic index (SDI), and geographic regions, using estimated annual percentage changes (EAPCs), joinpoint regression analysis, and Age-period-cohort (APC) analysis. The OA burden was decomposed into contributions from aging, population size, and epidemiologic changes, and the changes of OA burden through 2050 were predicted by Bayesian APC (BAPC) model.

Results

GBD 2021 estimated 607.0 million prevalent cases, 46.6 million incident cases, and 21.3 million YLDs of OA in 2021, with a globally increasing trend in overall burden observed from 1990 to 2021. The number of YLDs cases attributable to high BMI increased threefold, from 1.45 million in 1990 to 4.42 million in 2021. Joinpoint regression analysis indicated that the most pronounced increases were observed during 2006–2009 for prevalence, 2000–2009 for incidence, and 2006–2009 for YLDs. The risk of OA prevalence, incidence, and YLDs were influenced by age, period and cohort effect. Regarding the age effect, the relative risks for these outcomes initially increased and then declined. For the period effect, the relative risks increased by approximately 1.7-, 1.2-, and 1.7-fold for these outcomes. In the cohort effect, earlier-born cohorts demonstrated higher risks than their later-born counterparts. Results revealed that female individuals, knee OA, and middle SDI quintile were responsible for the most significant prevalence. The primary drivers of these changes were population growth and aging. Notably, the OA burden globally was predicted to increase from 2022 to 2050, with predicted values of 1.2 billion prevalent cases, 77.8 million incident cases, and 41.6 million YLDs by 2050.

Conclusion

As a major global health issue, OA demonstrated a steady rise in global burden between 1990 and 2021, highlighting the pressing need for improved OA management strategies. This study offers valuable insights for policymakers in optimizing healthcare resource distribution and designs targeted interventions.

The translational potential of this article

OA is a highly prevalent musculoskeletal disorder worldwide, with a globally increasing trend in its overall burden, imposing significant medical and economic challenges on both societies and individuals. Therefore, timely updates and comprehensive analyses of OA epidemiological data are crucial for the public, researchers, and healthcare policymakers. This study will contribute to the development of more effective OA management strategies and targeted interventions tailored to the specific needs and contexts of different regions and countries, with substantial translational potential.

Graphical abstract

1. Introduction

Osteoarthritis (OA) is a prevalent musculoskeletal condition mainly affecting middle-aged and elderly populations worldwide, primarily involving the hips, knees, feet, and hands [1]. Early OA typically presents with joint pain and reduced mobility, which may progress to significant functional decline or disability over time [2]. In 2019, global estimates included 527.8 million prevalent cases, 41.5 million incident cases, and 18.9 million years lived with disability (YLDs) attributed to OA [3]. These figures are projected to rise to 727.5 million prevalent cases, 52.9 million incident cases, and 26.0 million YLDs by 2035, driven by demographic aging and increasing obesity rate [3]. The lack of standardized diagnostic criteria for early OA, the limited evidence for non-surgical interventions, and inadequate public health awareness and healthcare system remain substantial obstacles to effective disease management and prevention. Therefore, given the high prevalence, incidence, YLDs and heavy burden of OA globally, to facilitate a more comprehensive global understanding of OA epidemiology, it is important to update the epidemiological data of OA for the public, researchers, health care policymakers.

While various studies have examined OA burden in the past decade, several gaps persist in existing literature (Table S1). For instance, some previous studies (1) focused on narrow demographic segments, such as working-age individuals [4] and postmenopausal women [5]; (2) only evaluated one metric of OA burden, such as prevalence [6] and disability-adjusted life years (DALYs) [7]; (3) only investigated the OA burden attributable to a specific risk factor, such as high body mass index (BMI) [8]; (4) investigated single country or region, such as China [9]; (5) addressed only specific OA subtypes, such as knee OA [10], hand OA [11], and hip OA [12]. Therefore, to address these limitations and advance global OA epidemiology understanding, this study provided the most up-to-date and comprehensive description of the OA epidemiology burden up to 2021 at the global, 21 GBD regional, and 204 national levels, including the three metrics of prevalence, incidence, and YLDs. The analyses were stratified by sex, joint site (including hand, hip, knee and an other), all age patterns, and socio-demographic index (SDI). We also assessed the contributions of high BMI as risk factors for OA and the inequalities in OA burden across different SDI levels. In addition, some previous studies lacked longitudinal trend analyses [13]. Therefore, we further analyzed the trends and primary drivers of OA burden using multiple statistical models and predicted the case number and ASR to 2050. Estimated annual percentage change (EAPC), joinpoint regression analysis, and age-period-cohort (APC) analysis were performed to evaluate the overall, local, and multidimensional trends in OA burden. Decomposition analysis was conducted to clarified the primary drivers of changes in OA burden. Finally, we projected the global OA burden to 2050 using the Bayesian APC (BAPC) model.

2. Methods

2.1. Data source

The Global Burden of Disease (GBD) Study 2021 utilized the latest available epidemiological data available and employed standardized, refined methodologies to comprehensively assess health loss associated with 371 diseases, injuries, and impairments [14,15]. This evaluation was conducted across 204 countries and territories, with stratification according to age and sex. To ensure a robust data integration framework, the study incorporated diverse data sources. For population-related data, the GBD 2021 study utilized a diverse range of sources, such as population censuses, vital registration systems, and national health surveys. Age- and sex-specific prevalence estimates were then produced using the DisMod-MR 2.1 model, a sophisticated statistical modeling tool. This model accounted for potential biases, thereby standardizing the data across different demographic profiles and geographical regions, which is crucial for ensuring the comparability and reliability of the results. OA and its subtypes were identified based on the ICD-10 codes. In the current investigation, we compiled data on the number of cases, as well as the age standardized rates (ASRs) of prevalence, incidence, and YLDs for OA. The data were stratified by sex, region, and country, covering the period from 1990 to 2021. The numerical estimates were presented alongside 95 % uncertainty intervals (95 % UIs), which quantify the level of uncertainty associated with each estimate. The rates were expressed per 100,000 individuals, accompanied by their corresponding 95 % UIs to provide a comprehensive understanding of the data variability. It is noteworthy that the GBD study does not account for OA-related mortality. Consequently, this analysis focused exclusively on non-fatal burden metrics, with a particular emphasis on YLDs, which offer insights into the years of healthy life lost due to the disease. High BMI (BMI ≥25 kg/m²) was identified as the primary and exclusive risk factor influencing OA in GBD database. To assess sociodemographic development, the study utilized the SDI. Based on their SDI values, the 204 countries and regions were categorized into five quintiles. This categorization allows for a nuanced analysis of the disease burden across different levels of sociodemographic development.

2.2. Descriptive analysis

To achieve a holistic comprehension of the burden imposed by OA, a descriptive analysis was performed at the global, regional, and national scales. A visual representation was generated to elucidate the global case counts and ASRs of OA prevalence, incidence, and YLDs for both overall and stratified by sex, spanning the period from 1990 to 2021. Meanwhile, a risk factor analysis was conducted to evaluate the effect of high BMI, the primary risk factor, on YLDs of OA. Furthermore, a comparative assessment was undertaken to examine the differences in the case numbers and ASRs of OA prevalence, incidence, and YLDs from 1990 to 2021 at multiple levels. This comparison encompassed the global level, 21 GBD geographical regions, 204 countries and territories, and five SDI-based quintiles.

2.3. Trend analysis

This study aimed to investigate OA trends from comprehensive, localized, and multidimensional viewpoints. To assess the overall trend of OA burden, the estimated annual percentage change (EAPC) was applied. A linear regression model was constructed as y = α+βx, with EAPC calculated as (exp(β)−1) × 100 %. In this model, y = ln(ASR) and x denotes the calendar year; the 95 % confidence interval (CI) was derived accordingly. An upward trend was defined when both the EAPC estimate and the lower bound of its 95 % CI exceeded 0. Conversely, a downward trend was defined when both the EAPC estimate and the upper bound of the 95 % CI were below zero.

To identify the local trends in OA burden, we utilized joinpoint regression analysis with the Joinpoint software (version 5.2.0; National Cancer Institute, Rockville, MD, US). This approach partitions the overall trend into distinct subintervals by detecting inflection points and quantifies each segment using the annual percentage change (APC) and its corresponding 95 % CI [[16], [17]]. Additionally, the average annual percentage change (AAPC) was calculated for the period spanning 1990 to 2021, with regression coefficients weighted by the duration of each interval. The AAPC and its 95 % CI were determined using the Monte Carlo permutation method, based on 4499 randomly permuted datasets. The overall significance level was adjusted using the Bonferroni correction. A specific period was considered to have an increasing trend if both the APC/AAPC estimate and the lower bound of its 95 % CI exceeded zero. Conversely, a decreasing trend was identified if both the APC/AAPC estimate and the upper bound of its 95 % CI were below 0. Otherwise, the trend was classified as stable.

In addition to analyzing the overall and local temporal trends of OA burden, we further employed an age-period-cohort (APC) model to investigate OA trends from multiple perspectives using Stata version 18. Because of the interaction among age, period, and birth cohort, it is difficult to determine their independent effects on the risk of prevalence, incidence and YLDs. Therefore, the APC model, which examines temporal changes of a certain variable by simultaneously incorporating three temporal dimensions, was applied to estimate the effects of these factors on the risk of the above-mentioned metrics. To address the multicollinearity between age, period, and birth cohort effects, we applied an advanced variant of the APC model using the intrinsic estimator (IE) method [18]. This method leverages principal component regression analysis to address the dynamic and heterogeneous influences of the age, period, and cohort dimensions. By doing so, it provides more efficient and stable estimates compared to traditional approaches. When constructed based on the Poisson distribution, the APC model employing the IE method can be mathematically represented as follows: ln(Y_i,j,k) = μ+α_i+β_j+γ_k+ε_i,j,k. In this equation, Y_i,j,k denotes the prevalence, incidence or YLDs of OA within the (i,j,k) group, which is defined by a specific age group (i), period group (j), and cohort group (k). The parameter μ represents the intercept of the model, serving as a baseline measure. The coefficients α_i, β_j, and γ_k respectively represent the impact of age, period, and cohort in their respective i-th, j-th, and k-th groups. The ε_i,j,k accounts for random error or residual variation. To facilitate the application of the APC model with the IE method, we re-coded the original data into consecutive 5-year age groups, ranging from 30 to 34 years to 95+ years. Similarly, we divided the study period from 1990 to 2021 into consecutive 5-year interval. Correspondingly, we generated consecutive 5-year birth cohort. This data recoding approach allowed for more precise estimation of the net effects of age, period, and cohort on OA prevalence, incidence, and YLDs. After applying the APC model using the IE method, we derived the estimated coefficients for these three factors. To improve the interpretability of these coefficients, we converted them into exponential values. This conversion allowed us to calculate the relative risks (RRs) for OA prevalence, incidence, and YLDs across specific age groups, period groups, or cohort groups, in comparison to the overall average level across all these three cohorts. These RRs provide valuable insights into the relative importance of each temporal factor in shaping the OA burden and can inform targeted public health interventions and preventive strategies.

2.4. Decomposition analysis

To uncover the drivers of changes in OA burden between 1990 and 2021, decomposition analyses based on the methodology of Das Gupta were performed across two dimensions: first, by sex and disease subgroups; and second, by age structure, population size, and epidemiological changes [5]. Initially, OA prevalence, incidence, and YLDs were stratified by disease categories and sexes. Subsequently, to assess the relative impacts of aging structure, population growth, and epidemiological shifts on the changes in OA epidemiology from 1990 to 2021, a decomposition analysis was conducted. This analysis dissected the changes in prevalence, incidence, and YLDs into components associated with age structure, population size, and age-standardized prevalent, incident, and YLDs rates (hereinafter termed epidemiological changes). The age-standardized rates were defined as the number of cases per 100,000 population. The Prevalence at each location can be calculated using the formula: Prevalence_ay,py,ey = ${\sum }_{i=1}^n\left(a_{i,y}\times p_y\times e_{i,y}\right)$ , where Prevalence_ay,py,ey represents the prevalence attributable to age structure, population size, and prevalent rate in year y; a_i,y denotes the proportion of the population in age category i out of n age categories in year y; p_y represents the total population in year y; and e_i,y indicates the prevalent rate for age category i in year y. The impact of each factor to the change in prevalence between 1990 and 2021 was evaluated by examining the effect of one factor at a time while keeping the other factors constant. For example, the effect of age structure was calculated as: [(Prevalence_{a2021, p1990, e1990} + Prevalence_{a2021, p2021, e2021})/3 + (Prevalence_{a2021, p1990, e2021} + Prevalence_{a2021, p2021, e1990})/6] - [(Prevalence_{a1990, p2021, e2021} + Prevalence_{a1990, p1990, e1990})/3 + (Prevalence_{a1990, p2021, e1990} + Prevalence_{a1990, p1990, e2021})/6]. Furthermore, the decomposition analysis was applied to incidence and YLDs.

2.5. Predictive analysis

While the preceding analyses focused on the historical burden of OA, a forward-looking projection of the OA burden is essential for informing public health policy and optimizing healthcare resource allocation. To achieve this, the Bayesian age-period-cohort (BAPC) model was employed in conjunction with the integrated nested Laplace approximation (INLA) to project the global OA burden through 2050. This method has demonstrated enhanced coverage and accuracy relative to conventional age-period-cohort models. As noted above, APC model could essentially be regarded as a multiple regression model, and its expression was as follows: ln(Y_i,j,k) = μ+α_i+β_j+γ_k+ε_i,j,k. Under the assumption that the effects of age, period, and cohort were similar across adjacent periods, Bayesian inference within APC models employs a second-order random walk to smooth the prior distributions for these effects and to predict posterior rates. By employing INLA within the BAPC framework, we were able to approximate marginal posterior distributions, efficiently bypassing several mixing and convergence issues commonly associated with the Markov Chain Monte Carlo sampling technique typically used in Bayesian analyses [19,20]. All analyses were performed using R software including the “BAPC” and “INLA” packages (version 4.4.2).

3. Results

3.1. Descriptive analysis of OA burden at global, regional, and national levels

In 2021, the Global number of prevalent cases of OA reached 606.99 million (95 % UI 537.87–670.52), corresponding to an ASR of 6967.29 per 100,000 individuals (95 % UI 6180.70–7686.06). This represents a 10.65 % increase between 1990 and 2021 (Table 1). In the same year, there were 46.63 million incident cases of OA globally (95 % UI 41.12–51.64), with an ASR of 535.00 per 100,000 people (95 % UI 472.38–591.97), marking a 9.03 % increase since 1990 (Table 2). Moreover, 21.30 million (95 % UI 10.19–42.94) YLDs were caused by OA worldwide in 2021, with an ASR of 244.50 (95 % UI 117.06–493.11), representing a 10.77 % increase from 1990 to 2021 (Table 3). Regionally, East Asia recorded the highest case numbers of prevalent, incident, and YLDs of OA, while the high-income Asia Pacific region exhibited the highest ASRs for all three metrics (Table 1, Table 2, Table 3). Nationally, China had the highest case numbers of OA prevalence, incidence, and YLDs, whereas the Republic of Korea reported the highest ASRs for these metrics (Fig. 1a; Fig. S1 and S2; Table S2). Regarding high BMI, the primary and exclusive risk factor for OA, 4.42 million (95 % UI -0.42–12.34) YLDs cases were attributed to it, corresponding to an ASR of 50.59 per 100,000 population (95 % UI -4.81–141.35), marking a 28.17 % increase form 1990 to 2021 (Table S3). In terms of SDI quintiles, the middle SDI quintile recorded the highest number of prevalent cases, incident cases, and YLDs, while the high SDI quintile exhibited the highest ASRs for prevalence, incidence, and YLDs in 2021 (Table 1, Table 2, Table 3).

Table 1.

Prevalent cases, age-standardized rate (ASR), percentage change and EAPC between 1990 and 2021 of osteoarthritis by sex, SDI quintiles and Global Burden of Disease (GBD) regions.

Area	Prevalent cases in 1990 (x 106)	Prevalent cases in 2021 (x 106)	Percentage change (1990–2021) (%)	ASR per 100,000 in 2021	Percentage change of ASR between 1990 and 2021 (%) (95 % UI)	EAPC (95 % CI)
Global	256.08 (227.12, 283.44)	606.99 (537.87, 670.52)	137.03	6967.29 (6180.70, 7686.06)	10.65 (9.98, 11.28)	0.34 (0.31, 0.38)
Sex
Male	99.51 (88.58, 110.78)	238.39 (211.38, 265.58)	139.55	5773.36 (5125.94, 6420.96)	11.38 (10.64, 12.07)	0.29 (0.27, 0.32)
Female	156.56 (138.99, 172.61)	368.60 (326.84, 407.39)	135.43	8049.41 (7137.25, 8892.72)	10.53 (9.84, 11.19)	0.38 (0.34, 0.43)
SDI
Low	11.82 (10.48, 13.24)	29.90 (26.52, 33.16)	152.97	5605.58 (4967.54, 6230.60)	11.76 (10.55, 13.08)	0.31 (0.29, 0.33)
Low-middle	33.53 (29.74, 37.22)	90.72 (80.47, 100.58)	170.54	6106.25 (5419.32, 6763.20)	16.86 (15.52, 18.16)	0.44 (0.42, 0.47)
Middle	65.08 (57.66, 72.19)	192.60 (170.45, 213.63)	195.94	6903.80 (6123.00, 7643.11)	16.55 (14.66, 18.32)	0.53 (0.49, 0.58)
High-middle	65.70 (57.97, 72.99)	140.87 (124.37, 156.19)	114.39	7120.38 (6297.95, 7879.76)	11.48 (10.24, 12.72)	0.37 (0.33, 0.43)
High	79.66 (71.44, 87.92)	152.39 (136.38, 168.14)	91.30	7897.27 (7067.13, 8689.88)	8.29 (7.66, 8.96)	0.25 (0.22, 0.30)
Central Europe, eastern Europe, and central Asia	33.32 (29.31, 37.38)	47.64 (41.88, 53.47)	42.99	7467.48 (6572.46, 8362.16)	8.39 (7.54, 9.25)	0.30 (0.26, 0.34)
Central Asia	2.89 (2.51, 3.29)	6.01 (5.21, 6.87)	107.95	7034.89 (6120.08, 8010.15)	16.36 (14.12, 18.50)	0.45 (0.39, 0.51)
Central Europe	9.38 (8.27, 10.51)	14.52 (12.79, 16.23)	54.79	6948.51 (6129.15, 7752.72)	13.20 (12.12, 14.35)	0.37 (0.36, 0.39)
Eastern Europe	21.05 (18.45, 23.75)	27.11 (23.77, 30.42)	28.80	7906.11 (6954.04, 8880.09)	6.49 (5.32, 7.64)	0.28 (0.22, 0.35)
High income	85.68 (76.98, 94.55)	156.40 (140.21, 172.57)	82.53	7874.86 (7075.90, 8675.66)	8.15 (7.50, 8.83)	0.25 (0.21, 0.29)
Australasia	1.66 (1.49, 1.83)	3.92 (3.54, 4.34)	136.32	7917.60 (7098.38, 8735.71)	9.71 (7.12, 12.49)	0.32 (0.30, 0.34)
High-income Asia Pacific	16.57 (14.69, 18.30)	34.63 (31.15, 37.98)	108.93	8608.63 (7674.07, 9485.19)	9.78 (8.29, 11.25)	0.54 (0.36, 0.74)
High-income North America	26.83 (24.15, 29.61)	51.75 (46.32, 57.32)	92.85	8421.62 (7534.98, 9282.03)	5.43 (4.47, 6.21)	0.02 (−0.11, 0.14)
Southern Latin America	3.25 (2.90, 3.60)	6.54 (5.89, 7.21)	101.31	7669.24 (6896.46, 8466.30)	11.49 (9.17, 14.01)	0.30 (0.27, 0.33)
Western Europe	37.37 (33.65, 41.32)	59.57 (53.85, 65.96)	59.40	7113.44 (6407.11, 7867.10)	6.76 (5.88, 7.73)	0.18 (0.15, 0.21)
Latin America and Caribbean	14.96 (13.24, 16.55)	46.87 (41.49, 51.72)	213.25	7426.34 (6579.41, 8178.83)	14.33 (13.49, 15.42)	0.40 (0.39, 0.40)
Andean Latin America	1.38 (1.23, 1.53)	4.43 (3.93, 4.89)	220.26	7370.44 (6552.10, 8123.13)	12.78 (10.64, 15.00)	0.35 (0.32, 0.37)
Caribbean	1.70 (1.50, 1.88)	3.85 (3.41, 4.25)	126.88	7134.56 (6327.26, 7876.59)	10.74 (9.07, 12.44)	0.33 (0.32, 0.34)
Central Latin America	5.70 (5.04, 6.30)	19.20 (16.95, 21.14)	236.93	7499.49 (6635.38, 8259.93)	15.48 (14.15, 17.00)	0.40 (0.40, 0.41)
Tropical Latin America	6.18 (5.48, 6.85)	19.39 (17.18, 21.55)	213.58	7424.65 (6582.79, 8241.00)	13.95 (12.78, 15.19)	0.41 (0.40, 0.41)
North Africa and Middle East	9.34 (8.29, 10.40)	30.49 (27.06, 33.71)	226.54	6265.22 (5572.94, 6946.23)	18.64 (17.05, 20.23)	0.47 (0.43, 0.51)
South Asia	32.45 (28.77, 35.90)	96.53 (85.58, 106.69)	197.43	6326.13 (5612.39, 7009.64)	19.08 (17.37, 20.76)	0.53 (0.50, 0.56)
Southeast Asia, east Asia, and Oceania	68.35 (60.02, 76.46)	198.02 (174.65, 221.05)	189.73	6704.51 (5927.12, 7455.97)	18.70 (16.64, 20.82)	0.60 (0.53, 0.68)
East Asia	55.51 (48.47, 62.11)	158.29 (139.47, 176.82)	185.17	7036.10 (6216.29, 7835.76)	18.90 (16.67, 21.07)	0.62 (0.54, 0.71)
Oceania	0.18 (0.16, 0.20)	0.51 (0.45, 0.57)	187.61	6196.48 (5474.55, 6895.02)	11.04 (8.56, 13.54)	0.28 (0.25, 0.31)
Southeast Asia	12.66 (11.24, 14.12)	39.23 (34.57, 43.61)	209.79	5675.80 (5001.76, 6320.89)	21.79 (19.64, 23.93)	0.56 (0.55, 0.57)
Sub-Saharan Africa	11.97 (10.62, 13.37)	31.03 (27.57, 34.38)	159.16	6107.38 (5411.38, 6782.17)	12.05 (11.01, 13.16)	0.33 (0.32, 0.34)
Central Sub-Saharan Africa	1.31 (1.16, 1.46)	3.52 (3.12, 3.91)	167.67	5940.49 (5268.27, 6589.54)	6.54 (4.29, 8.53)	0.10 (0.04, 0.15)
Eastern Sub-Saharan Africa	3.91 (3.47, 4.38)	10.41 (9.27, 11.56)	166.05	5829.96 (5160.63, 6476.62)	15.60 (13.55, 17.53)	0.44 (0.42, 0.45)
Southern Sub-Saharan Africa	1.80 (1.59, 2.00)	4.29 (3.77, 4.76)	138.04	7161.23 (6333.34, 7951.33)	11.27 (10.10, 12.51)	0.31 (0.30, 0.33)
Western Sub-Saharan Africa	4.94 (4.38, 5.53)	12.81 (11.40, 14.17)	159.14	6075.81 (5385.72, 6757.27)	11.82 (10.84, 12.84)	0.33 (0.32, 0.35)

Table 2.

Incident cases, age-standardized rate (ASR), percentage change and EAPC between 1990 and 2021 of osteoarthritis by sex, SDI quintiles and Global Burden of Disease (GBD) regions.

Area	Incident cases in 1990 (x 106)	Incident cases in 2021 (x 106)	Percentage change (1990–2021) (%)	ASR per 100,000 in 2021	Percentage change of ASR between 1990 and 2021 (%) (95 % UI)	EAPC (95 % CI)
Global	20.90 (18.47, 23.10)	46.63 (41.12, 51.64)	123.11	535.00 (472.38, 591.97)	9.03 (7.56, 10.56)	0.33 (0.31, 0.35)
Sex
Male	8.50 (7.50, 9.45)	18.93 (16.69, 21.02)	122.60	445.74 (393.68, 493.98)	9.45 (7.99, 11.03)	0.28 (0.25, 0.30)
Female	12.40 (10.97, 13.67)	27.70 (24.48, 30.57)	123.47	621.32 (548.49, 686.94)	8.57 (7.12, 10.07)	0.36 (0.32, 0.39)
SDI
Low	1.09 (0.97, 1.22)	2.80 (2.48, 3.12)	156.25	447.12 (395.36, 493.37)	11.70 (9.62, 13.71)	0.30 (0.28, 0.32)
Low-middle	3.04 (2.68, 3.38)	7.81 (6.89, 8.68)	157.40	480.13 (424.96, 533.01)	16.83 (14.40, 19.51)	0.41 (0.39, 0.45)
Middle	5.75 (5.06, 6.41)	15.44 (13.57, 17.14)	168.44	536.49 (473.16, 595.02)	13.51 (11.42, 15.64)	0.41 (0.39, 0.45)
High-middle	5.13 (4.54, 5.68)	10.35 (9.10, 11.47)	101.61	548.07 (481.66, 608.49)	9.75 (8.39, 11.20)	0.40 (0.38, 0.43)
High	5.86 (5.23, 6.48)	10.19 (9.09, 11.27)	73.74	611.30 (542.71, 675.91)	8.47 (7.50, 9.44)	0.24 (0.22, 0.24)
Central Europe, eastern Europe, and central Asia	2.42 (2.13, 2.71)	3.27 (2.89, 3.66)	35.29	552.97 (487.52, 616.65)	1.36 (−0.22, 2.71)	0.30 (0.27, 0.32)
Central Asia	0.22 (0.19, 0.25)	0.48 (0.41, 0.54)	114.48	504.46 (442.21, 565.75)	8.01 (4.10, 11.77)	0.42 (0.37, 0.46)
Central Europe	0.70 (0.62, 0.78)	0.96 (0.85, 1.07)	37.30	522.05 (460.81, 580.35)	3.05 (1.25, 4.90)	0.34 (0.34, 0.36)
Eastern Europe	1.50 (1.31, 1.68)	1.83 (1.61, 2.06)	22.59	584.97 (515.25, 651.42)	−2.82 (−4.74, −0.97)	0.29 (0.25, 0.34)
High income	6.25 (5.58, 6.93)	10.32 (9.21, 11.46)	65.02	614.07 (545.98, 679.78)	8.72 (7.77, 9.68)	0.21 (0.18, 0.24)
Australasia	0.12 (0.11, 0.14)	0.27 (0.24, 0.30)	116.97	620.09 (550.76, 686.53)	12.97 (9.38, 16.67)	0.35 (0.32, 0.39)
High-income Asia Pacific	1.35 (1.19, 1.49)	2.19 (1.96, 2.41)	62.69	682.07 (606.06, 752.84)	6.19 (4.61, 7.83)	0.45 (0.30, 0.58)
High-income North America	1.90 (1.70, 2.09)	3.46 (3.06, 3.85)	81.99	646.38 (572.29, 715.37)	12.38 (10.59, 14.21)	−0.01 (−0.16, 0.09)
Southern Latin America	0.25 (0.22, 0.28)	0.48 (0.43, 0.54)	91.56	596.27 (530.39, 660.42)	13.99 (10.53, 17.50)	0.31 (0.27, 0.35)
Western Europe	2.63 (2.35, 2.93)	3.92 (3.50, 4.36)	48.96	557.66 (497.27, 618.53)	5.13 (3.89, 6.54)	0.20 (0.17, 0.23)
Latin America and Caribbean	1.32 (1.17, 1.47)	3.79 (3.35, 4.19)	186.77	585.62 (517.63, 647.75)	15.41 (13.66, 17.01)	0.37 (0.37, 0.38)
Andean Latin America	0.12 (0.11, 0.13)	0.36 (0.32, 0.40)	199.99	578.18 (511.33, 641.10)	19.69 (16.08, 23.22)	0.35 (0.33, 0.37)
Caribbean	0.14 (0.12, 0.15)	0.30 (0.26, 0.33)	112.91	555.77 (493.24, 617.51)	6.53 (3.54, 9.45)	0.30 (0.29, 0.31)
Central Latin America	0.51 (0.45, 0.57)	1.56 (1.38, 1.73)	206.41	589.49 (521.45, 652.52)	16.09 (13.85, 18.41)	0.37 (0.37, 0.38)
Tropical Latin America	0.55 (0.49, 0.61)	1.57 (1.39, 1.73)	184.34	589.12 (521.39, 650.60)	14.70 (12.44, 17.07)	0.39 (0.38, 0.39)
North Africa and Middle East	0.84 (0.74, 0.93)	2.73 (2.40, 3.04)	225.39	488.31 (433.70, 542.33)	14.98 (12.07, 17.81)	0.43 (0.41, 0.45)
South Asia	2.99 (2.64, 3.33)	8.22 (7.24, 9.12)	174.63	495.01 (436.64, 548.03)	16.47 (13.59, 19.62)	0.46 (0.43, 0.48)
Southeast Asia, east Asia, and Oceania	5.98 (5.24, 6.69)	15.36 (13.49, 17.24)	156.77	522.89 (459.30, 584.39)	17.51 (15.18, 20.10)	0.55 (0.50, 0.61)
East Asia	4.84 (4.23, 5.42)	12.05 (10.56, 13.56)	149.17	554.47 (486.91, 619.37)	23.13 (20.53, 26.05)	0.59 (0.51, 0.68)
Oceania	0.02 (0.01, 0.02)	0.05 (0.04, 0.05)	186.59	480.95 (423.12, 536.36)	8.37 (2.47, 15.15)	0.25 (0.22, 0.28)
Southeast Asia	1.13 (0.99, 1.26)	3.26 (2.87, 3.64)	188.90	437.13 (386.13, 485.01)	21.96 (19.22, 24.64)	0.51 (0.50, 0.51)
Sub-Saharan Africa	1.09 (0.96, 1.22)	2.94 (2.59, 3.27)	168.83	481.51 (425.43, 532.13)	15.28 (13.24, 17.23)	0.30 (0.30, 0.31)
Central Sub-Saharan Africa	0.12 (0.11, 0.14)	0.34 (0.30, 0.38)	178.43	463.08 (409.68, 515.24)	3.90 (−0.09, 7.79)	0.09 (0.05, 0.14)
Eastern Sub-Saharan Africa	0.37 (0.32, 0.41)	1.00 (0.89, 1.12)	173.01	461.02 (407.42, 509.93)	14.48 (12.21, 16.97)	0.38 (0.37, 0.39)
Southern Sub-Saharan Africa	0.16 (0.14, 0.18)	0.38 (0.33, 0.42)	138.12	557.24 (493.49, 618.18)	13.60 (11.06, 16.28)	0.29 (0.28, 0.30)
Western Sub-Saharan Africa	0.45 (0.40, 0.50)	1.23 (1.08, 1.37)	173.60	483.84 (427.08, 536.68)	19.62 (17.01, 22.21)	0.34 (0.33, 0.34)

Table 3.

Years lived with disability (YLDs) cases, age-standardized rate (ASR), percentage change and EAPC between 1990 and 2021 of osteoarthritis by sex, SDI quintiles and Global Burden of Disease (GBD) regions.

Area	YLDs cases in 1990 (x 106)	YLDs cases in 2021 (x 106)	Percentage change (1990–2021) (%)	ASR per 100,000 in 2021	Percentage change of ASR between 1990 and 2021 (%) (95 % UI)	EAPC (95 % CI)
Global	8.92 (4.26, 17.98)	21.30 (10.19, 42.94)	138.87	244.50 (117.06, 493.11)	10.77 (8.01, 12.55)	0.19 (0.17, 0.22)
Sex
Male	3.43 (1.64, 6.91)	8.28 (3.96, 16.67)	141.24	200.52 (95.88, 404.87)	13.03 (10.32, 14.98)	0.31 (0.29, 0.34)
Female	5.49 (2.62, 11.08)	13.02 (6.24, 26.27)	137.39	284.14 (136.29, 573.11)	9.61 (6.54, 11.49)	0.42 (0.37, 0.47)
SDI
Low	0.40 (0.19, 0.80)	1.02 (0.49, 2.04)	155.64	190.93 (91.63, 384.26)	20.11 (15.71, 23.25)	0.45 (0.44, 0.46)
Low-middle	1.14 (0.55, 2.28)	3.11 (1.49, 6.25)	173.91	209.35 (100.40, 422.62)	21.58 (17.55, 24.60)	0.51 (0.47, 0.55)
Middle	2.24 (1.08, 4.49)	6.71 (3.20, 13.47)	200.09	240.41 (115.09, 483.99)	16.18 (13.16, 18.83)	0.60 (0.58, 0.63)
High-middle	2.29 (1.09, 4.62)	4.96 (2.37, 9.97)	116.27	250.58 (119.78, 503.69)	12.97 (10.29, 15.19)	0.34 (0.32, 0.35)
High	2.85 (1.37, 5.72)	5.49 (2.65, 11.05)	92.85	283.13 (136.04, 570.53)	2.62 (0.59, 4.46)	0.34 (0.31, 0.36)
Central Europe, eastern Europe, and central Asia	1.17 (0.56, 2.38)	1.69 (0.82, 3.42)	44.56	264.68 (127.04, 534.59)	9.24 (5.27, 11.79)	0.34 (0.31, 0.37)
Central Asia	0.10 (0.05, 0.20)	0.21 (0.10, 0.43)	109.49	249.29 (119.63, 500.56)	16.32 (11.81, 20.21)	0.61 (0.61, 0.62)
Central Europe	0.33 (0.16, 0.66)	0.51 (0.25, 1.04)	57.32	245.41 (117.63, 496.21)	15.32 (11.16, 18.42)	0.30 (0.28, 0.33)
Eastern Europe	0.74 (0.36, 1.51)	0.97 (0.47, 1.95)	30.07	280.78 (134.37, 567.03)	5.95 (1.41, 9.07)	0.51 (0.30, 0.72)
High income	3.06 (1.47, 6.16)	5.64 (2.73, 11.38)	84.13	282.65 (136.22, 569.11)	2.06 (−0.01, 3.97)	0.31 (0.28, 0.34)
Australasia	0.06 (0.03, 0.12)	0.14 (0.07, 0.29)	139.61	283.38 (139.24, 577.97)	13.34 (9.69, 17.79)	0.66 (0.58, 0.75)
High-income Asia Pacific	0.60 (0.29, 1.21)	1.28 (0.61, 2.58)	113.71	314.98 (150.55, 636.77)	10.36 (8.30, 12.47)	0.43 (0.43, 0.44)
High-income North America	0.96 (0.46, 1.95)	1.86 (0.90, 3.76)	92.60	300.89 (144.87, 606.97)	−7.32 (−9.97, −4.64)	0.07 (−0.08, 0.21)
Southern Latin America	0.12 (0.05, 0.23)	0.23 (0.11, 0.47)	102.91	273.53 (131.22, 548.73)	7.82 (3.83, 11.31)	0.39 (0.37, 0.40)
Western Europe	1.33 (0.64, 2.67)	2.13 (1.04, 4.29)	60.57	253.58 (123.06, 510.55)	5.10 (2.97, 6.89)	0.37 (0.35, 0.39)
Latin America and Caribbean	0.52 (0.25, 1.05)	1.65 (0.79, 3.33)	217.24	261.20 (124.97, 527.98)	14.28 (10.76, 17.14)	0.38 (0.36, 0.40)
Andean Latin America	0.05 (0.02, 0.10)	0.16 (0.08, 0.32)	223.12	260.94 (125.19, 526.82)	13.74 (8.67, 18.86)	0.65 (0.58, 0.71)
Caribbean	0.06 (0.03, 0.12)	0.14 (0.06, 0.27)	127.73	251.06 (120.09, 508.25)	5.86 (3.45, 8.67)	0.37 (0.33, 0.40)
Central Latin America	0.20 (0.09, 0.40)	0.68 (0.32, 1.37)	241.91	264.58 (126.15, 535.65)	15.68 (11.82, 18.76)	0.50 (0.46, 0.55)
Tropical Latin America	0.21 (0.10, 0.43)	0.68 (0.33, 1.37)	218.00	259.93 (124.74, 524.63)	14.38 (10.31, 17.68)	0.40 (0.36, 0.44)
North Africa and Middle East	0.32 (0.15, 0.64)	1.05 (0.50, 2.12)	228.51	215.92 (103.37, 437.62)	15.69 (12.25, 18.84)	0.45 (0.43, 0.46)
South Asia	1.09 (0.53, 2.20)	3.31 (1.58, 6.66)	202.56	216.90 (104.00, 438.04)	26.00 (21.43, 29.54)	0.17 (0.12, 0.23)
Southeast Asia, east Asia, and Oceania	2.34 (1.13, 4.72)	6.89 (3.29, 13.81)	194.30	233.18 (111.66, 467.78)	21.77 (19.00, 24.82)	0.30 (0.29, 0.32)
East Asia	1.90 (0.92, 3.83)	5.52 (2.63, 11.06)	189.79	245.04 (117.45, 492.41)	22.45 (19.75, 25.74)	0.32 (0.28, 0.37)
Oceania	0.01 (0.00, 0.01)	0.02 (0.01, 0.03)	189.12	212.87 (102.99, 428.52)	12.28 (7.96, 16.82)	0.51 (0.49, 0.52)
Southeast Asia	0.43 (0.21, 0.87)	1.36 (0.65, 2.71)	214.22	196.18 (93.56, 393.42)	26.44 (22.42, 30.25)	0.38 (0.37, 0.39)
Sub-Saharan Africa	0.41 (0.20, 0.82)	1.07 (0.51, 2.15)	161.44	211.11 (101.16, 425.95)	18.30 (13.92, 21.35)	0.50 (0.48, 0.53)
Central Sub-Saharan Africa	0.04 (0.02, 0.09)	0.12 (0.06, 0.24)	169.73	204.30 (97.78, 412.23)	17.22 (11.92, 22.52)	0.41 (0.39, 0.43)
Eastern Sub-Saharan Africa	0.13 (0.06, 0.27)	0.36 (0.17, 0.72)	169.45	200.97 (96.11, 405.27)	24.60 (18.74, 28.82)	0.27 (0.21, 0.33)
Southern Sub-Saharan Africa	0.06 (0.03, 0.13)	0.15 (0.07, 0.30)	137.09	249.45 (120.36, 500.63)	−0.44 (−4.10, 2.77)	0.27 (0.21, 0.33)
Western Sub-Saharan Africa	0.17 (0.08, 0.34)	0.44 (0.21, 0.89)	162.00	210.09 (101.11, 424.67)	20.01 (16.09, 23.13)	0.57 (0.53, 0.61)

Fig. 1.

Global Age-standardized rate (ASR) of prevalence. (a) ASR of prevalence of osteoarthritis (OA) per 100,000 by country for both sexes combined in 2021. (b) Global prevalence ASR of total and different site of osteoarthritis (OA) in 2021 by sex and age.

3.2. Sex, joint site, and age patterns of OA burden

In 2021, the global counts and ASRs of OA prevalence, incidence, and YLDs were higher among women than men across all age groups. This sex-related pattern was observed in hand and knee OA but was not evident in hip and other types of OA (Fig. 1b; Fig. S3 and S4; Table S4–6). For both sexes, the ASRs of OA prevalence and YLDs increased with age. However, in knee OA, these metrics rose with age and peaked at 80–84 years for both sexes (Fig. 1b; Fig. S4). The ASR of OA incidence reached its highest level in the 55–59 years age group and subsequently showed a fluctuating decline, with a slight uptick in the 95+ years age group (Fig. S3). In terms of anatomic sites, the knee was the predominant site for OA across all age groups in 2021. It had the highest number of prevalent cases (374.74 million), incident cases (30.85 million), and YLDs (12.02 million) globally. Additionally, the knee OA had the highest ASRs of prevalence (4294.27 per 100,000 people), incidence (353.67 per 100,000 people), and YLDs (137.59 per 100,000 people). The knee accounted for the largest proportion of global OA burden in 2021, with 56.18 % of prevalent cases, 66.15 % of incident cases, and 56.42 % of YLDs, followed by the hand (29.12 %, 22.23 %, and 28.95 %, respectively), the other (9.32 %, 7.77 %, and 9.28 %) and the hip (5.38 %, 3.85 %, and 5.35 %). Moreover, the knee remained the primary site of OA across most SDI categories and GBD regions, whereas the hand was the leading site in Eastern Europe and Central Asia for prevalence and YLDs in 2021 (Figs. S5a, S6a, S7a, S8; Table S7–11). The prevalence and YLDs of OA increased with age for both sexes, reaching a peak in the 65–69 years age group. In contrast, the incidence of OA increased with age and peaked in the 50–54 years age group for both sexes (Figs. S5b, S6b, S7b).

3.3. Overall trends in OA burden

Between 1990 and 2021, the global ASRs of OA prevalence, incidence, and YLDs increased by an average of 0.34 %, 0.33 %, and 0.19 % per year, respectively (Table 1, Table 2, Table 3). Regionally, East Asia reported the highest EAPC for prevalence and incidence, while Australasia showed the highest EAPC for YLDs. Conversely, High-income North America exhibited the lowest EAPC for prevalence, incidence, and YLDs (Table 1, Table 2, Table 3). Nationally, Equatorial Guinea experienced the largest increases in the ASRs of prevalence, incidence, and YLDs, whereas Israel experienced the greatest decreases in the ASRs of all three metrics (Table S2). Additionally, within the SDI quintiles, the middle SDI quintile exhibited the highest EAPC for prevalence and YLDs, both the middle and low-middle SDI quintiles showed the highest EAPC in incidence, while the high SDI quintile showed the lowest EAPC for these three metrics (Table 1, Table 2, Table 3).

3.4. OA burden by socio-demographic index

Overall, a positive correlation was detected between the ASR of OA prevalence and the SDI from 1990 to 2021 (Fig. 2a). Notably, at the regional level, certain regions, particularly High-income Asia Pacific and High-income North America, did not exhibit this positive correlation. In Southeast Asia, North Africa and the Middle East, Central Europe, and Western Europe, the observed ASR of OA prevalence fell below the SDI-based expected levels during 1990–2021 (Fig. 2a). Similarly, in 2021, positive associations were obseved between SDI and the ASRs of OA prevalence, incidence, and YLDs across 204 countries and territories. In numerous countries and territories, such as the United States, Japan, Russia, and Brazil, the ASRs of all three metrics exceeded the expected level (Fig. 2b, Fig. S9).

Fig. 2.

OA burden by socio-demographic index. (a) Age-standardized rate (ASR) of prevalence of global and 21 Global Burden of Disease (GBD) regions of osteoarthritis (OA) by sociodemographic index (SDI) from 1990 to 2021. The expected values, based on the relationship between rate of OA prevalence and SDI across global and 21 GBD regions, are represented by the black line. For each GBD region, 32 data points are plotted, representing the observed age-standardized years lived with disability (YLDs) rates from 1990 to 2021. (b) Age-standardized prevalence rates of OA by SDI for 204 countries and territories in 2021. The expected values are indicated by the blue line, and each point represents the observed ASR of prevalence for a specific country or territory.

3.5. Joinpoint regression analysis on OA burden

The joinpoint regression analysis of global cases and the ASRs of prevalence, incidence, and YLDs of OA is depicted in Fig. 3. From 1990 to 2021, the case numbers for prevalence, incidence, and YLDs of OA consistently and significantly increased. The most rapid increases occurred during the periods of 2006–2009 for prevalence, 2000–2009 for incidence, and 2006–2009 for YLDs, following five joinpoints for each metric (Table S12). Regarding the ASRs of prevalence, incidence, and YLDs, generally mild upward trends were observed. The most rapid increases were seen during 2006–2009 for prevalence and YLDs, and during 2000–2005 for incidence. However, for the ASR trends of prevalence and YLDs, there was no change from 2015 to 2018, followed by a noticeable increase from 2018 to 2021 (Table S13).

Fig. 3.

The joinpoint regression analysis on the global case numbers and the age-standardized rates (ASRs) of prevalence, incidence and years lived with disability (YLDs) of osteoarthritis (OA).

3.6. Age-period-cohort analysis on OA burden

Results of age-period-cohort (APC) analyses of the OA prevalence, incidence, and YLDs were presented in Fig. S10. After adjusting for period and cohort effects, the age effect showed a significant influence on the risk of OA prevalence, incidence, and YLDs. The relative risks of OA prevalence, incidence, and YLDs all exhibited an initial increase followed by a decline, peaking at ages 70–74, 55–59, and 70–74 years, respectively. Notably, the relative risk of incidence exhibited a secondary increase among individuals aged 90–95 years (Table S14). After adjusting for age and cohort effects, the period effect demonstrated a notable influence on the risks of OA prevalence, incidence, and YLDs, and exhibited a subtle upward trend over time. Specifically, the RRs for these outcomes increased by 1.7-, 1.2-, and 1.7-fold, respectively, between 1992 and 2017. The most elevated risks for prevalence, incidence, and YLDs were recorded during the 2017 period, as presented in Table S14. Following adjustment for age and period effects, the birth cohort effect emerged as a substantial determinant of the risk associated with OA prevalence, incidence, and YLDs. Notably, earlier-born cohorts exhibited a greater risk of prevalence, incidence, and YLDs compared to their later-born counterparts. The RRs for these outcomes showed a consistent decline across birth cohorts, starting from the 1897–1901 cohort and progressing to the 1987–1991 cohort, as illustrated in Table S14.

3.7. Decomposition analysis on OA prevalence, incidence and YLDs

Between 1990 and 2021, a substantial surge in the global prevalence, incidence, and YLDs of the disease under study was observed. The most significant increase was identified within the middle SDI quintile, as depicted in Fig. 4 and Figures S11, S12. An analysis of the factors contributing to the global rise in prevalence revealed that the aging structure, population size, and epidemiological shifts accounted for 15.83 %, 73.77 %, and 10.4 % of the increase, respectively. Similarly, for incidence, the aging structure, population size, and epidemiological shifts were responsible for 9.13 %, 79.5 %, and 11.36 % of the global increase, respectively. Regarding YLDs, the aging structure, population size, and epidemiological shifts contributed 15.79 %, 73.1 %, and 11.11 %, respectively, to the global increase. The contributions of these factors varied across different SDI quintiles. In the high SDI quintile, population aging contributed the most significantly to the increase, accounting for 28.5 %, 15.31 %, and 28.44 % of the total change for prevalence, incidence, and YLDs. Population growth played a dominant role in the low SDI quintile, accounting for 93.69 %, 92.30 %, and 92.60 % of the rise for these three metrics. Epidemiological changes had the most substantial effect on prevalence and YLDs in the low-middle SDI quintile, and on incidence in the high-middle SDI quintile, with a contribution of 14.46 %, 15.55 %, and 13.63 % (Table S15–17). When the data were stratified by disease categories and sexes, it became evident that the influences of demographic and epidemiological factors on prevalence, incidence, and YLDs varied across subgroups. Knee OA accounted for the highest burden across disease categories, contributing 61.25 %, 64.95 %, and 55.5 % of total prevalence, incidence, and YLDs, respectively. Females exhibited a higher burden than males in all disease categories except for the subgroup of other OA. Apparently, population growth exerted the most pronounced influence on prevalence, incidence, and YLDs across all SDI quintiles and subgroups. Notably, the negative effects of aging structure on prevalence, incidence, and YLDs were observed exclusively in the low SDI quintile, with a more pronounced presentation in males than females (Fig. 4; Fig. S11 and S12; Table S15–17).

Fig. 4.

Changes in prevalent cases of osteoarthritis (OA) from 1990 to 2021, attributed to aging structure, population growth, and epidemiological changes, are shown at the global and across Sociodemographic Index (SDI) quintiles level, stratified by disease categories and sex. The black dot indicates the total value of change in OA prevalence resulting from the combined effects of all three components. A positive value for a given component reflects an increase in OA prevalence attributable to that component, while a negative value indicates a decrease.

3.8. Predictive analysis on OA burden to 2050

The projected case counts and ASRs for the prevalence, incidence, and YLDs of OA through 2050 are presented in Fig. 5 and Fig. S13. On a global scale, the model predicts an upward trend in the case numbers of prevalence, incidence, and YLDs for both males and females. By 2050, the projected values are 1.2 billion prevalent cases, 77.8 million incident cases, and 41.6 million YLDs. With regard to the ASRs, it is anticipated that the ASRs for prevalence and YLDs will increase annually. In contrast, the ASR for incidence is expected to follow a different trajectory, initially rising and then undergoing a slight decrease by 2050. The detailed projected values for the case numbers and ASRs of prevalence, incidence, and YLDs of OA are outlined in Table S18 and S19.

Fig. 5.

Prediction of (a) the number of prevalent cases, (b) age-standardized rate (ASR) of prevalence, (c) the number of incident cases, (d) ASR of incidence, (e) the number of YLDs, and (f) ASR of YLDs for osteoarthritis (OA) globally by sex from 1990 to 2050. Solid lines represent the observed values (1990–2021), while dashed lines indicate the predicted values (2022–2050).

4. Discussion

Results from the current study confirm that OA represents a substantial and growing public health concern. The global burden of OA, as measured by prevalence, incidence, and YLDs, increased steadily from 1990 to 2021. This upward trend is particularly concerning given the projected demographic shifts, including population growth and aging, which are expected to continue driving the burden of OA upward in the coming decades. This indicates that the overall impact of OA on global health will remain significant. In terms of policy implications, our findings underscore the necessity of continued efforts to improve OA management and prevention strategies. Given the projected increase in OA burden over the coming decades, it is essential to invest in research and development of new treatments, prevention strategies, and healthcare delivery models. This includes the development of early diagnostic criteria for OA, which could facilitate earlier intervention and better management of the disease [9]. Additionally, efforts to promote healthy lifestyles, including regular physical activity, weight management, and joint protection, could help reduce the risk of OA and improve outcomes for those affected by the disease [7,21].

Results of this study underscore the notable differences in the burden of OA based on sex, joint site, and age. The sex differences observed in our study are noteworthy. Women consistently exhibited higher rates of OA prevalence, incidence, and YLDs compared to men across all age groups and anatomical sites. This finding is consistent with prior research showing that women are at higher risk of developing OA, particularly in the hand and knee joints. The reasons for this sex difference are not fully understood but may include differences in hormonal factors, reproductive history, body composition, menopause, and joint mechanics [22,23]. Furthermore, the sex-specific disparities call for gender-sensitive approaches in OA care, including early screening for women at high risk (e.g., postmenopausal individuals) and tailored rehabilitation programs [24]. Further research is warranted to elucidate the mechanisms underlying these sex differences and to develop targeted interventions to reduce the OA burden in women. Additionally, the knee emerged as the most affected joint, accounting for the highest number and ASRs of prevalence, incidence, and YLDs. These results underscore the substantial public health impact of knee OA, which remains a leading cause of disability worldwide [25]. Given its association with obesity, sedentary lifestyles, and occupational strain, public health interventions should prioritize weight management, physical activity promotion, and ergonomic workplace modifications [10,26,27]. Age-related trends revealed that OA burden escalates with advancing age, peaking in different life stages depending on the metric. This discrepancy suggests that while new OA diagnoses are most frequent in midlife, the cumulative disability and chronicity of the disease intensify in later decades [13].

The regional and national variations in OA burden observed in our study highlight the importance of considering local context in efforts to address the OA burden. While some regions, such as East Asia, consistently exhibit high levels of OA burden, others, such as high-income Asia Pacific, display higher ASRs despite having lower absolute case numbers. These variations may reflect differences in demographic structure, healthcare access, lifestyle factors, and environmental exposures [6,28]. The East Asian region comprises countries including China, Japan, South Korea, North Korea, and Mongolia. Among these, China accounts for the largest case numbers of OA cases, primarily due to its large population base, relatively slower socioeconomic development, limited healthcare accessibility, and lower levels of mechanization, all of which have contributed to the rapid increase in OA cases and exerted a substantial influence on the regional estimates. In contrast, high-income Asia Pacific countries such as South Korea and Japan have smaller absolute populations and case numbers. However, the severity of population aging in these countries has resulted in relatively high age-standardized rates (ASRs). Therefore, public health interventions to address the OA burden should be adapted to the specific needs and contexts of individual regions and countries. In addition, high SDI countries consistently shoulder a disproportionately high share of the OA burden. This finding is consistent with previous studies showing that wealthier nations often experience higher rates of chronic diseases, driven by factors including longer life expectancy, sedentary lifestyles, and higher rates of obesity [12,29]. However, our study also reveals that the association between the SDI and OA burden is not uniform across all regions. For instance, while high-income Asia Pacific and high-income North America exhibit the highest ASRs, some regions such as Southeast Asia, North Africa and the Middle East, Central Europe, and Western Europe display ASRs that are lower than expected based on their SDI levels. This suggests that other factors, such as cultural practices, healthcare accessibility, and environmental exposures, may also play important roles in shaping the OA burden in different regions.

The joinpoint analysis conducted in this study segmented the overall OA burden into distinct time periods, offering a unique perspective for understanding the temporal trends of OA burden. Notably, during the final stage of the study period (2018–2021), the ASRs of prevalence and YLDs of OA showed a marked increased trend. This notable increase may be attributed to several factors, especially the potential impact of the COVID-19 pandemic [14]. As a major public health crisis, COVID-19 may act as a novel risk factor for OA, potentially accelerating its occurrence and development through both direct and indirect pathways. SARS-CoV-2, the virus responsible for COVID-19, has been reported to directly invade joint tissues, resulting in substantial joint damage. In one study, two patients with COVID-19 exhibited rapid joint destruction, characterized by cystic lesions at the osteochondral junction. These findings were further corroborated in a golden Syrian hamster model, which reproduced the observed joint destruction [30]. Meanwhile, COVID-19 infection induced a systemic inflammatory response, characterized by elevated levels of pro-inflammatory cytokines. These cytokines, rather than the virus itself, could infiltrate joint tissues, leading to synovitis and subsequent joint damage [31]. Additionally, lifestyle changes, behavioral shifts, psychological stress, and healthcare system disruptions caused by the COVID-19 pandemic may have contributed to the rising burden of OA [32]. COVID-19 pandemic led to reduced physical activity, resulting in muscle atrophy and deminished joint stability, and contributing to the development of OA. Additionally, pandemic related stress and changes in dietary habits contributed to weight gain, which imposed additional strain on weight-bearing joints, such as the knees and hips, accelerating the progression of OA [33]. The COVID-19 pandemic diverted healthcare resources toward addressing the acute needs of infected patients, which resulted in delays in the diagnosis and treatment of chronic conditions such as OA. This disruption of care may have contributed to worsening osteoarthritis symptoms and accelerated disease progression in many patients [34,35].

The age-period-cohort analysis conducted in this study provides a nuanced understanding of the temporal trends in OA burden. The age effect demonstrated a clear pattern of rising and then falling risk of OA prevalence, incidence, and YLDs, with the highest risks observed in older age groups, likely due to cumulative joint damage over time [36]. However, the age effect was not uniform across all age groups, with some subgroups exhibiting higher risks at younger ages. This suggests that there may be specific age-related risk factors or protective factors that warrant further investigation. The relative risks of OA incidence reached the highest levels at 55–59 years. This increasing risk could be attributed to multiple factors, including overweight and obesity driven by poor dietary habits and insufficient physical activity, exercise-related injuries, and the growing prevalence of unhealthy lifestyles among young and middle-aged individuals, underscoring the urgent need for targeted preventive strategies. The relative risks of OA prevalence and YLDs reached the highest levels at 70–74 years, significantly later than the incidence. This discrepancy suggests that while new OA diagnoses are most frequent in midlife, the cumulative disability and chronicity of the disease intensify in later decades. The period effect, which reflects the influence of environmental and societal factors impacting the entire population, also demonstrated a significant impact on OA burden. The slightly increasing trends in period effects for prevalence, incidence, and YLDs suggest that there may be broader societal or environmental changes that are influencing the OA burden. For instance, the global obesity epidemic, which associated with an increased risk of OA, may be contributing to the observed period effects [36]. Additionally, changes in occupational patterns, physical activity levels, and healthcare access may also be influencing OA burden over time [37]. With period progressing, the gradual improvement of health registration systems may have partially contributed to the observed increase in OA burden. The birth cohort effect, which reflects changes in the risk of OA in children born in similar years due to shared life events, also demonstrated a significant influence on OA burden. The elevated OA risk observed in earlier birth cohorts compared to more recent cohorts suggests that there may be cohort-specific risk factors or protective factors that are influencing OA burden. For instance, earlier birth cohorts may have been exposed to higher levels of physical labor or occupational hazards, which could increase their risk of developing OA later in life. Alternatively, later birth cohorts may benefit from improvements in healthcare access, nutrition, and living conditions, which could reduce their risk of OA [4].

The decomposition analysis conducted in our study provides important insights into the drivers of changes in OA burden over time. Population growth and aging were identified as the major contributors to the increase in OA prevalence, incidence, and YLDs globally. This finding underscores the importance of addressing demographic shifts in efforts to mitigate the OA burden [38]. In particular, the aging of the global population is expected to have a profound impact on OA prevalence, as older adults are at higher risk of developing OA due to cumulative joint wear and tear over time [39]. Moreover, when stratified by SDI, the impact of population growth on OA burden was most pronounced in the low SDI quintiles, emphasizing the need for region-specific interventions to manage the increasing demand for OA-related healthcare services. When stratified by joint site, the patterns of the knee, hip, and other OA were consistent with those of overall OA, indicating that population aging contributed more to the OA burden than epidemiological changes. In contrast, the hand OA exhibited the opposite trend, with epidemiological changes emerging as the secondary contributing factors for OA burden, particularly at the global and middle SDI levels. This discrepancy can be attributed to the hand joint is less affected by aging than weight-bearing joints such as knee and hip. When stratified by sex, the driving patterns of OA burden were generally consistent, although the overall burden was substantially higher in female than in male. Interestingly, our study also reveals that epidemiological changes, while contributing to the increase in OA burden to a lesser extent than population growth and aging, still play a significant role. This suggests that improvements in OA diagnosis, reporting, and management practices over time may have led to an increase in the recorded burden of OA. However, it is also possible that changes in risk factors, such as obesity rates and physical activity levels, have contributed to the observed epidemiological changes. Further research is needed to disentangle these effects and identify specific risk factors that can be targeted for intervention [40].

The number cases of global OA prevalence, incidence, and YLDs that predicted by BAPC analysis will increased by 97.2 %, 66.8 %, and 95.4 % from 2021 to 2050. This substantial increase was primarily attributable to population growth, particularly the expansion of the elderly population, and was strikingly consistent with the findings of the decomposition analysis, which should be highlighted in the formulation of OA related public health policies and the planning of intervention strategies.

This study has several limitations to consider. First, the exclusion of individuals aged 0–29, along with inconsistencies in data collection methods and quality, may have introduced potential bias. Furthermore, the GBD study employed the DisMod-MR 2.1 model, which accounts for potential biases. Data obtained through diverse diagnostic criteria were adjusted to mitigate systematic bias, a process referred to as crosswalking. However, this would further lead to a heavy reliance on the estimated data, especially at the national level. Therefore, the use of independent national registry data for verification may help address the limitation of data availability in certain countries. The exclusion of individuals aged 0–29 may have led to an underestimation of incidence risk among younger populations. Nevertheless, given the relatively low incidence in this group and the limited number of medical visits, the overall impact on this study is considered negligible. In addition, although this study examined the impact of high BMI, the only attributable burden quantified risk factor for OA in GBD study, on OA YLDs, other potential risk factors, including injury history, occupational exposures, sedentary behavior, aging, and socioeconomic status, were not further analyzed owing to data limitations. In decomposition analysis, each factor's contribution to changes in the OA burden was evaluated independently, assuming that the other factors remained constant. However, this approach inevitably ignores potential interactions among factors. This limitation may lead to an underestimation of the overall effect of composite factors on the OA burden. The timeliness of this study largely depends on updates to the GBD database, as the release of new data will inevitably necessitate revisions of the descriptive results derived from the OA burden estimates. Nevertheless, the results from multiple modeling approaches used in this study, including joinpoint regression analysis, APC analysis and BAPC predictive analysis retain a certain degree of credibility and stability, which can be cross-validated with future GBD data and provide a theoretical basis for their interpretation. Moreover, continuous updates of the GBD database will further help verify the robustness of our predictive findings. Additional efforts are required to address these limitations.

5. Conclusion

In conclusion, our study provides a comprehensive and up-to-date assessment of the global burden of OA, including detailed analyses of trends, decomposition, and projections to 2050. Our findings reveal that OA represents a substantial and growing public health concern. These findings underscore the urgent need for more effective OA management strategies and targeted interventions to address the specific needs and contexts of different regions and countries. Investing in research and development, promoting healthy lifestyles, and tackling the disparities in OA burden across countries are essential steps towards reducing the OA burden worldwide and improving outcomes for those affected by the disease.

Author Contributors

Study conception and design: GX, ZS and MC. Study conduct and data collection: MC, SC, CX, FL, CT, WG, MQ and SL. Data analysis: MC, SC, CX, ZS and GX. Data interpretation: GX, ZS, MC and SC. Drafting the manuscript: GX, MC and SC. GX and ZS take the responsibility for the integrity of the data analyses and approve the final version of the manuscript.

Declaration of generative AI and AI-assisted technologies in the writing process

During the preparation of this work the author(s) used ChatGPT-4o in order to improve grammar and check spelling. After using this tool, the author(s) reviewed and edited the content as needed and take(s) full responsibility for the content of the publication.

Data availability statement

Data used for the analyses are publicly available from https://ghdx.healthdata.org/gbd-2021.

Author notes

This study is based on publicly available data and solely reflects the opinion of its authors and not that of the Institute for Health Metrics and Evaluation.

Funding

This work was supported in part by the Shenzhen Medical Research Funds (B2504003, B2402033), the National Natural Science Foundation of China (Grants 82302767, 81803917, 81974352, 82022047, 82250710175, 82261160395, 82430078, 82230081, 82004395, 82402788), the Hubei Provincial Natural Science Foundation of China (Grants 2024AFB610 and 2024AFB651), the Scientific Research Project of Hubei Provincial Administration of Traditional Chinese Medicine (Grant ZY2025M052), and the Guangdong Provincial Science and Technology Innovation Council Grant (2017B030301018).

Declaration of competing interest

The authors declare there is no competing interest.

Appendix A. Supplementary data

The following are the Supplementary data to this article.