Global variation in the quality of care for rheumatoid arthritis and associated factors

Abstract

Background

Although the quality of care related to rheumatoid arthritis (RA) has improved, ensuring high-quality care globally remains a significant challenge. To address this issue, we have introduced a modified Quality of Care Index (QCI) to evaluate variations in RA care services worldwide and analyze the influencing factors.

Methods

The QCI was derived from a principal component analysis of global incidence, mortality, and prognostic indicators of RA. Joinpoint regression and linear mixed models were employed to analyze the temporal trends of the QCI and its influencing factors.

Result

In 2021, the global QCI for RA was 72.09. Among this, the QCI for males was 77.25, while for females it was 71.12. Based on Joinpoint regression, the AAPC of the global RA QCI from 1990 to 2021 was 0.30(0.29-0.31), with 0.22(0.20-0.23) for males and 0.29(0.28-0.30) for females. Based on the LMM model, it was found that age, gender, year, and SDI were all statistically significantly associated with QCI (p < 0.05). Specifically, positive correlations with QCI were observed in the following groups: under 14 years, 20–24 years, 40–54 years, 70–74 years, males, and high-SDI regions. Conversely, negative correlations with QCI were identified in the age groups 15–19 years, 25–39 years, 55–69 years, and 75 years and above.

Conclusion

Disparities in RA-related care exist across gender, age, and geographic regions. Further emphasis should be placed on improving care for female RA patients and those in low-SDI regions.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13075-026-03737-6.

Introduction

Rheumatoid arthritis (RA) is an autoimmune disease characterized by synovitis as its pathological basis, which may eventually lead to joint deformity [1]. Its etiology involves autoimmune dysregulation, genetic susceptibility, and dysbiotic microbial triggers, although no single factor is sufficient to cause disease [2]. The clinical manifestations of RA primarily involve joint pathology, with initial symptoms often presenting as morning stiffness, swelling, and pain in the joints [3]. As the disease progresses, it may develop into joint deformities, impairing normal joint function [4]. Some patients may also exhibit systemic symptoms, such as fever, fatigue, and weakness [5]. The treatment of rheumatoid arthritis largely depends on the specific condition of the patient, with conventional approaches including pharmacotherapy and surgical intervention [6]. Although there is currently no complete cure for the disease, early intervention and sustained maintenance therapy can effectively slow disease progression and improve patients' quality of life [7].

RA can develop at any age, with women having an incidence rate 2–3 times higher than men. The peak onset occurs between 40 and 60 years of age, and the disease leads to a significant health burden [8]. According to the Global Burden of Disease (GBD) database, there were 17.6 million cases of RA globally in 2021, and 38,300 deaths were attributed to the disease [9]. Effective treatment and care are crucial for improving RA prognosis [10]. Non-pharmacological management of RA should include patient education, shared decision-making, exercise, orthotic devices, and multidisciplinary care approaches [11]. Pharmacological treatment should involve conventional synthetic disease-modifying antirheumatic drugs (DMARDs), with methotrexate as the first-line option [12]. If monotherapy with a DMARD fails to achieve treatment targets, combination therapy with DMARDs, biological DMARDs, and targeted synthetic DMARDs should be considered [13]. RA management should also encompass monitoring, pre-treatment investigations, vaccinations, as well as screening for tuberculosis and hepatitis [14]. If non-surgical treatments prove ineffective, surgical intervention may be recommended [15]. These measures collectively form the comprehensive care framework for RA [16].

By understanding the risk factors, mechanisms, and consequences of RA and implementing evidence-based prevention measures, the prognosis of RA can be significantly improved [17]. From a quality-of-care perspective, it is crucial to globally evaluate and strengthen response strategies to this significant issue and identify risk factors affecting the quality of disease management [18]. To conduct a robust multidimensional assessment of RA-related care quality, we adopted a more comprehensive and refined Quality of Care Index (QCI) [19]. This approach aims to address gaps identified in previous evaluations, deepen the understanding of factors influencing care quality, and thereby inform future interventions and policy decisions.

Methods

Data sources

Data on the global burden of RA were sourced from the GBD database. The metrics extracted included incidence, prevalence, mortality, disability-adjusted life years (DALYs), years lived with disability (YLDs), and years of life lost (YLLs). These estimates are disaggregated by age and encompass global totals, regions categorized by the Socio-demographic Index (SDI), as well as 204 countries and territories. The complete dataset is accessible through the GBD results tool.

Quality of care index

To evaluate disease burden and quality of care, we calculated four key ratios using six core metrics sourced. These indices include: the prevalence-to-incidence ratio (PIR), the mortality-to-incidence ratio (MIR), the years of life lost to years lived with disability ratio (YLR), and the disability-adjusted life years to prevalence ratio (DPR). These were calculated using the following formulas:

The PIR measures the ratio between prevalent and incident cases. A high PIR suggests a chronic disease with prolonged duration, while a low PIR is typical of acute diseases that resolve quickly or progress rapidly [20]. The MIR reflects disease lethality by measuring the proportion of new cases that lead to death. Therefore, it serves as an indicator of healthcare quality, with lower values corresponding to better treatment outcomes [21]. The YLR evaluates the relative weight of fatal versus non-fatal burden—higher values indicate a predominance of premature mortality, whereas lower values point to disability as the main component of burden [22]. The DPR estimates the average health loss per prevalent case in relation to how common the disease is. Diseases that are severe but rare tend to have a high DPR, while common diseases with mild health impacts per case exhibit a low DPR [23].

The computed PIR, MIR, YLR, and DPR indices first underwent standardization, whereby each variable was scaled to a mean of zero and a standard deviation of one. Principal component analysis (PCA) was subsequently applied to the normalized dataset [24]. The advantage of the PCA method lies in its ability to handle multicollinearity among multiple correlated disease burden indicators by transforming them into uncorrelated principal components. This approach ensures objectivity in deriving weights based on data variance, thereby avoiding subjective assignment [25]. From the PCA output, the rotation matrix corresponding to the first two principal components was extracted, and weight coefficients (ω₁ to ω₈) were assigned based on their respective loadings. These weights were then used to compute scores for PC1 and PC2. Here, the coefficients ω₁–ω₈ are the PCA-derived loadings that quantify how much each ratio contributes to its respective component. PC1 embodies the dominant source of variation in the data, while PC2 accounts for residual variability left unexplained by PC1. In general, the first principal component (PC1) accounts for the largest share of variance within the dataset, representing the dominant pattern of variability. In contrast, the second principal component (PC2) was primarily influenced by the PIR metric. The specific formulas are:

PC1 represents the principal axis of variation in the data, while PC2 captures residual variability not explained by PC1. The PCA score is calculated using the formula:

Finally, the QCI index was calculated using the formula:

The code used for this computation has been uploaded as supplementary materials.

Joinpoint regression

We employed the Joinpoint Regression Program (version 5.1.0.0), a specialized tool for trend analysis, to analyze the age-standardized QCI of RA. This software uses joinpoint regression to identify the most parsimonious model that fits the data, allowing users to specify the range of joinpoints to test [26]. From this analysis, we derived the Average Annual Percentage Change (AAPC, %) along with its 95% confidence interval (CI).

Linear mixed models

We employed a linear mixed model (LMM) to evaluate the effects of multiple factors on the RA-related QCI, while adjusting for potential cross-cutting confounders including age, gender, year, and SDI region [27]. The LMM was chosen due to its capacity to accommodate hierarchical data structures and account for inherent correlations in the dataset. To address non-independence in estimates resulting from repeated measurements across geographic regions, we specified location as a random effect. This approach effectively accounts for within-region clustering, producing more robust and reliable parameter estimates. Prior to LMM analysis, along with the categorical variables—age, gender, and year—both SDI and QCI values were standardized.

Results

QCI of RA

In 2021, the global QCI for RA was 72.09. Among this, the QCI for males was 77.25, while for females it was 71.12. The QCI across different age groups and SDI regions globally in 2021 is presented in Figure 1. The age-standardized QCI for various countries and territories worldwide is shown in Figure 2. Table S1-S3 in the appendix(supplementary files) displays the age-standardized YLR, DPR, MIR, PIR, and QCI for different countries and territories globally in 2021.

Fig. 1.

The QCI across different age groups (A) and SDI regions (B) globally in 2021

Fig. 2.

The age-standardized QCI for various countries and territories worldwide in 2021. Note: A Both genders; B Male; C Female

Joinpoint regression

Based on Joinpoint regression, the AAPC of the global RA QCI from 1990 to 2021 was 0.30(0.29–0.31.29.31), with 0.22(0.20–0.23.20.23) for males and 0.29(0.28–0.30.28.30) for females. The AAPCs for global and different SDI regions are shown in Table 1 and Figure 3.

Table 1.

Temporal trends in QCI of RA across Global and SDI regions

location	Gender	AAPC.Index	P.Value
Global	Both	0.30(0.29–0.31.29.31)	<0.05
Global	Female	0.29(0.28–0.30.28.30)	<0.05
Global	Male	0.22(0.20–0.23.20.23)	<0.05
High SDI	Both	0.33(0.32–0.34.32.34)	<0.05
High SDI	Female	0.28(0.26–0.29.26.29)	<0.05
High SDI	Male	0.28(0.25–0.31.25.31)	<0.05
High-middle SDI	Both	0.34(0.32–0.36.32.36)	<0.05
High-middle SDI	Female	0.34(0.33–0.36.33.36)	<0.05
High-middle SDI	Male	0.19(0.17–0.21.17.21)	<0.05
Middle SDI	Both	0.44(0.42–0.45.42.45)	<0.05
Middle SDI	Female	0.47(0.45–0.48.45.48)	<0.05
Middle SDI	Male	0.41(0.38–0.43.38.43)	<0.05
Low-middle SDI	Both	0.23(0.21–0.26.21.26)	<0.05
Low-middle SDI	Female	0.29(0.26–0.31.26.31)	<0.05
Low-middle SDI	Male	0.25(0.22–0.28.22.28)	<0.05
Low SDI	Both	0.13(0.11–0.16.11.16)	<0.05
Low SDI	Female	0.20(0.17–0.23.17.23)	<0.05
Low SDI	Male	0.07(0.04–0.10.04.10)	<0.05

Fig. 3.

Temporal trends in QCI of RA across Global and SDI regions

Linear mixed models

Based on the LMM model, it was found that age, gender, year, and SDI were all statistically significantly associated with QCI (p < 0.05). Specifically, positive correlations with QCI were observed in the following groups: under 14 years, 20–24 years, 40–54 years, 70–74 years, males, and high-SDI regions. Conversely, negative correlations with QCI were identified in the age groups 15–19 years, 25–39 years, 55–69 years, and 75 years and above. The effects of these different factors on QCI are illustrated in Figure 4 and detailed in Table 2. The relationship between QCIs in different SDI regions around the world is shown in Figure 5A. The top and bottom ten countries of QCI are presented in Figure 5B.

Fig. 4.

Factors Influencing QCI and Their Effects. Notes: A Effects of ages. B Effects of SDI. C Effects of gender. D Effects of year. E Effects of age and gender. F Effects of year and SDI

Table 2.

Analysis of variance for fixed effects

Factors	Sum Sq	Mean Sq	NumDF	DenDF	F value	Pr(>F)
Age	376,801.80	19,831.67	19.00	433,665.56	177,149.93	<0.05
Gender	27.07	13.53	2.00	433,665.56	120.89	<0.05
Year	2.19	2.19	1.00	5,505.42	19.60	<0.05
SDI	12.78	12.78	1.00	3,884.33	114.17	<0.05

Fig. 5.

The relationship between QCIs in different SDI regions around the world. Notes: A QCIs in different SDI regions around the world. B The top and bottom ten countries

Discussion

We initially employed the QCI to assess the global landscape of RA care in 2021. The QCI is a standardized composite indicator that enables direct comparison of care quality across countries and regions [23]. This facilitates the identification of disparities and the monitoring of progress over time—information essential for guiding resource allocation and designing targeted health interventions.The findings suggest that a foundational framework for RA management has been established worldwide, likely associated with the widespread adoption of DMARDs such as methotrexate [28]. Nevertheless, the overall quality remains substantially below the desired standard. While certain high-SDI nations achieved QCI scores exceeding 90, some African countries recorded scores below 30.

Moreover, a gender-based disparity in QCI was observed in 2021, with male RA patients demonstrating higher scores than their female counterparts. Women inherently experience a higher incidence of RA, often presenting with more complex symptomatology that can be mistaken for other conditions, resulting in delayed diagnosis [29]. Prior to confirmation, the disease frequently causes greater irreversible joint damage, thereby compromising the baseline for care quality [30]. Familial caregiving responsibilities and socioeconomic factors may further predispose women to postpone seeking medical attention [31]. In low- and middle-income regions characterized by limited healthcare resources and pronounced gender inequality, this disparity is markedly exacerbated [32].

It is noteworthy that female RA patients are more susceptible to chronic pain and functional impairment, necessitating long-term quality-of-life management strategies [33]. Enhancing global RA care quality fundamentally depends on addressing this gender gap [34]. Future initiatives must prioritize the implementation of early RA screening programs targeted at women to reduce diagnostic delays [35]. Additionally, greater emphasis should be placed on chronic disease management, pain control, functional rehabilitation, and quality-of-life improvements of RA [36]. In underserved regions, investments in primary healthcare infrastructure should be coupled with targeted efforts to eliminate cultural and economic barriers that impede women's access to medical services [37].

To understand the changes in the QCI for RA over the past 30 years, we conducted an analysis using Joinpoint regression. The results indicate that the QCI for RA has shown signs of improvement globally over the last three decades, including in both high and low SDI regions. However, the extent of improvement in high-SDI regions has been more substantial compared to that in low-SDI regions. In recent years, breakthrough advancements have been made in the treatment of rheumatoid arthritis [38]. Inhibiting the release of inflammatory cytokines by blocking the JAK-STAT signaling pathway can rapidly alleviate symptoms, proving particularly effective for patients who do not respond to traditional medications [39]. Monoclonal antibodies targeting specific inflammatory factors such as IL-6 and TNF-α, as well as CD20 monoclonal antibodies that target B cells, allow for precise regulation of abnormal immune responses and reduce joint damage [40]. S1P receptor modulators mitigate inflammation by inhibiting lymphocyte migration and exhibit fewer side effects compared to conventional drugs [41]. The combined detection of anti-CCP antibodies and rheumatoid factor has improved early diagnosis rates [42]. Additionally, testing for genetic polymorphisms in RA patients helps predict differential drug responses, reducing the costs associated with trial-and-error treatment strategies [43]. However, these technologies require significant healthcare expenditures, making it difficult for them to be rapidly adopted in low-SDI regions [44].

We employed a Linear Mixed Model to conduct an in-depth analysis of the key factors influencing the QCI for RA. The model results indicated that age, gender, year, and SDI all demonstrated statistically significant associations with QCI (p < 0.05). The LMM first revealed a non-linear, wave-like relationship between age and QCI, rather than a simple monotonic change with increasing age. Juvenile Rheumatoid Arthritis is a chronic autoimmune disease primarily characterized by synovitis in childhood and accompanied by impairments in multiple organ functions, mostly occurs in children aged 4–16 [45]. The course of this condition can extend for several years, with alternating acute episodes and remissions [46]. Most cases resolve spontaneously by adulthood, although a small number may continue to have active disease [47]. It is noteworthy that the QCI for elderly RA patients is relatively low, particularly for those over 75 years old. This is likely because the focus of treatment shifts towards safety and comorbidity management rather than aggressive disease control, while issues such as frailty and cognitive impairment can also impact the quality of care [48].

Based on the LMM, we also found that the QCI values for male patients were significantly higher than those for females. This gap further reveals significant gender inequality in RA care. Over time, the QCI values showed a stable upward trend. This further demonstrates that globally, the overall quality of RA care has achieved significant and continuous improvement over the past 30 years. The LMM also confirmed a clear positive relationship between SDI level and predicted QCI values. Higher SDI is associated with a higher QCI. High-SDI regions possess more abundant medical resources, a more robust healthcare security system, and greater public health awareness [49]. This means patients in these regions have easier access to early diagnosis, advanced treatments, and continuous management, significantly enhancing the quality of care [50]. Consequently, some new and expensive treatment modalities are often prioritized for use in high-SDI regions. Beyond considering the promotion of new technologies in low-SDI regions, employing artificial intelligence to bridge the QCI gap between different SDI regions could be a viable strategy [51]. AI-assisted diagnosis for RA is considered feasible [52]. Using AI to predict treatment efficacy for RA is also deemed meaningful [53].

This study has several limitations. As the GBD estimates are model-derived, disparities in raw data quality among regions and countries may introduce bias into the source datasets. This could affect the accuracy of true RA burden estimates—for example, resulting in underreporting in low- and middle-SDI countries due to insufficient infrastructure and health policy frameworks, or overestimation in countries where data are primarily sourced from urban areas.

Conclusion

Disparities in RA-related care exist across gender, age, and geographic regions. Future efforts must prioritize closing the gender gap—especially for women in low-SDI settings—through early-diagnosis programmes, subsidised access to advanced DMARDs, and community-based multidisciplinary care.

Abbreviations

RA

Rheumatoid Arthritis

QCI

Quality of Care Index

CI

Confidence interval

GBD

Global Burden of Disease

SDI

Socio-Demographic Index

AAPC

Annual Percentage Change

ICD

International Classification of Diseases

DALYs

Disability-Adjusted Life Years

YLDs

Years Lived with Disability

YLLs

Years of Life Lost

LMM

Linear Mixed Models

MIR

Mortality to Incidence Ratio

PIR

Prevalence to Incidence Ratio

YLR

Years Lived with Disability Ratio

DPR

Disability Adjusted Life years to Prevalence Ratio

Authors’ contributions

Study design: DZS and LXC. Data collection: LXC. Data analyses: FKB. Results interpretations: LXC. Manuscript writing: FKB and HXX. Manuscript proofing: DZS. Correspondence to Xiaocong Lin and Kaibin Fang.

Funding

This study is funded by Fujian Provincial Science and Technology Innovation Joint Project Plan (No. 2024Y9349), Fujian provincial health technology project (No. 2020CXA045) and the Research Fund for PhD Tutorship of the Second Affiliated Hospital of Fujian Medical University(No. 2022BD0301).

Data availability

To download the data used in these analyses, please visit the Global Health Data Exchange GBD 2021 data-input sources tool at https://vizhub.healthdata.org/gbd-results/?params=gbd-api-2021-public. No permission is required for anyone to access this data.

Declarations

Ethics approval and consent to participate

This study utilized anonymized data compiled by the Institute for Health Metrics and Evaluation (IHME) at the University of Washington. The research protocol, including the waiver of informed consent, was reviewed and approved by the Institutional Review Board (IRB) of the University of Washington.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.