Risk factors and predictive model for mild cognitive impairment in elderly patients with rheumatoid arthritis

Abstract

Background

Rheumatoid arthritis (RA) is a chronic autoimmune disorder characterized by joint destruction and systemic inflammation, both of which significantly impair patients’ quality of life. Mild cognitive impairment (MCI), a reversible precursor to dementia, is increasingly prevalent among elderly RA patients. Early identification of MCI in this population allows for timely interventions to slow cognitive decline.

Objective

This study aims to identify independent risk factors for MCI in elderly patients with RA and to develop a predictive nomogram.

Methods

We enrolled 378 elderly RA patients, aged 60 to 80 years, from Xi’an Fifth Hospital between December 2023 and December 2024. Cognitive function was assessed using the Montreal Cognitive Assessment (MoCA), with scores ranging from 20 to 26 indicating MCI. We analyzed demographic, clinical, and laboratory data to identify risk factors through logistic regression and constructed a nomogram. The model’s performance was evaluated using receiver operating characteristic (ROC) curves, calibration plots, and decision curve analysis (DCA).

Results

Among the 378 patients, 94 (24.87%) were classified in the RA-MCI group. Multivariate analysis identified the course of disease (COD) (OR = 1.07, 95% CI: 1.03–1.10), elevated Disease Activity Score-28 (DAS28) (OR = 1.31, 95% CI: 1.13–1.53), high C-reactive protein (CRP) levels (OR = 1.01, 95% CI: 1.01–1.02), and osteoporosis (OP) (OR = 1.88, 95% CI: 1.14–3.13) as independent risk factors. The nomogram demonstrated moderate discrimination (AUC = 0.750, 95% CI: 0.696–0.805) and clinical utility.

Conclusion

The COD, OP, DAS28, and CRP levels are key predictors of MCI in elderly RA patients. The proposed nomogram provides a practical tool for early risk stratification, facilitating targeted interventions to delay cognitive decline.

Trial registration

This study conformed to the principles outlined in the Declaration of Helsinki and received approval from the Medical Ethics Committee of Xi’an Fifth Hospital (Approval No.: [2023] Ethics Review 55). Additionally, the trial was registered with the Chinese Clinical Trial Registry (Registration No.: ChiCTR2300077337, Registration Date: 2023-11-01). Written informed consent was obtained from all individual participants included in the study.

Introduction

Rheumatoid arthritis (RA) is an autoimmune disease characterized by erosive synovitis and joint destruction, primarily affecting young to middle-aged individuals and imposing a significant burden on both patients and society [1]. Extra-articular manifestations of RA are closely associated with disease activity, while systemic inflammation notably increases the risk of cardiovascular and cerebrovascular events, contributing to elevated mortality among RA patients [2]. Emerging evidence suggests that both localized and systemic inflammation in RA may contribute to cognitive decline and even dementia, with cognitive impairment becoming a key determinant of disability in this population [3].

Dementia, a progressive neurodegenerative disorder, remains poorly understood in terms of its pathogenesis, and current therapeutic approaches show limited efficacy [4]. Recent studies have identified systemic inflammation as a key contributor to the pathogenesis of Alzheimer’s disease (AD), wherein inflammatory cells release mediators that induce neuronal degeneration and cognitive decline [5]. Microglia and astrocytes, which are critical regulators of neuroinflammation in the central nervous system, can disrupt brain structure and function when hyperactivated [6]. Macrophage migration inhibitory factor, a cytokine vital for immune regulation, plays a dual role in RA by recruiting inflammatory mediators and activating neuroinflammatory responses derived from microglia or astrocytes, thus contributing to cognitive dysfunction in RA patients [7].

Mild cognitive impairment (MCI), recognized as a transitional state between normal aging and dementia, is prevalent among older adults and significantly increases the risk of progression to dementia [8]. Studies report an annual conversion rate of 10–15% from MCI to dementia, with this rate progressively rising to 80% within six years [9]. A cohort study by the California Alzheimer’s Disease Research Center, which involved 327 MCI patients, found that 65% progressed to dementia and 24% died during a three-year follow-up period [10]. As a prodromal stage of dementia, MCI shares multiple risk factors with dementia, including advanced age, female sex, and low educational attainment [11]. Targeted screening, diagnosis, and intervention in high-risk MCI populations may substantially reduce dementia incidence [12]. The Montreal Cognitive Assessment (MoCA), a widely used tool for evaluating cognitive function, exhibits superior sensitivity in detecting early cognitive deficits in MCI and AD compared to the Mini-Mental State Examination (MMSE), particularly in RA patients [13].

This study employed the MoCA to evaluate cognitive function, developed an early diagnostic risk model for rheumatoid arthritis-related MCI (RA-MCI), and constructed a nomogram with internal validation to predict the risk of cognitive decline in elderly RA patients. These efforts aim to provide novel insights into early diagnosis and intervention strategies for cognitive impairment in this vulnerable population.

Methods

Study population

Elderly RA patients who visited the Rheumatology Department of Xi’an Fifth Hospital between December 1, 2023, and December 1, 2024, were enrolled in the study cohort. Inclusion Criteria:(1) Meeting the 2010 American College of Rheumatology/European League Against Rheumatism (ACR/EULAR) classification criteria for rheumatoid arthritis [14]; (2) Aged 60 to 80 years; (3)MoCA scores between 20 and 30; (4) Availability of reliable inpatient or outpatient medical records; (5) Ability to complete neuropsychological assessments; (6) Signed informed consent. Exclusion Criteria: (1) Diagnosis of other autoimmune disorders; (2) Severe comorbidities (e.g., hepatic or renal dysfunction, myocardial infarction, malnutrition, respiratory disorders, hematologic diseases); (3) History of cognitive disorders (e.g., Alzheimer’s disease, cerebral hemorrhage, vascular dementia, or post-stroke sequelae); (4) Alcohol abuse or substance dependence. All patients provided signed informed consent.

From December 1, 2023, to December 1, 2024, a total of 929 RA patients were admitted to the Rheumatology Department of Xi’an Fifth Hospital. Among these, 524 patients were under 60 years of age, 10 patients were over 80 years, and 395 patients met the age criteria of 60 to 80 years. Initially, 395 elderly RA patients were enrolled; however, after applying exclusion criteria, 17 patients were excluded from the study: 2 patients with severe pulmonary disease, 4 patients with comorbid immune disorders, 1 patient with acute heart failure, 2 patients with post-stroke sequelae, 1 patient with a history of cerebral hemorrhage, 1 patient with renal failure, 1 patient with chronic benzodiazepine use, and 5 patients with MoCA scores of 20 or lower. Consequently, 378 patients were included in the analysis. Based on the MoCA scores, patients were classified into the RA-MCI group and the non-RA-MCI group. A patient enrollment flowchart is presented in Fig. 1.

Fig. 1.

Research flowchart

Ethical considerations

This study adhered to the principles outlined in the Declaration of Helsinki and was approved by the Medical Ethics Committee of Xi’an Fifth Hospital (Approval No.: [2023] Ethics Review 55). Furthermore, the trial was registered with the Chinese Clinical Trial Registry (Registration No.: ChiCTR2300077337, Registration Date: 2023-11-01).

Data collection

Demographic and clinical data were collected from enrolled patients, including gender, age, COD, DAS28, and serum biomarkers. The serum biomarkers comprised erythrocyte sedimentation rate (ESR), measured using the Westergren method; C-reactive protein (CRP) and rheumatoid factor (RF), quantified via rate nephelometry; and anti-cyclic citrullinated peptide (anti-CCP) antibody levels, assessed by microparticle enzyme immunoassay.

Cognitive function assessment

Cognitive function was evaluated using the MoCA, which assesses several domains, including visuospatial/executive function, naming, memory, language, abstract thinking, and orientation. Participants with 12 or fewer years of formal education received an additional point. The total score ranges from 0 to 30 [15].

Statistical analysis

This study aims to identify predictors of RA-MCI and develop a predictive model. According to the international guidelines for predictive modeling (TRIPOD) [16], the sample size in the training set should be at least ten times the number of predictors. Based on previous studies [17] and the TRIPOD principles, a total of 378 elderly RA patients were included (94 in the RA-MCI group and 284 in the non-RA-MCI group). Patient data, including gender, age, COD, DAS28, osteoporosis (OP), CRP, ESR, RF, and anti-CCP, were collected. The sample size met the requirement (at least 90 events for 9 predictive variables), ensuring sufficient statistical power for the model.

Data were analyzed using SPSS version 27.0 as follows: Normally distributed continuous variables were expressed as mean ± standard deviation (SD) and compared using Student’s t-test. Non-normally distributed variables were reported as median [interquartile range (IQR)] and analyzed using the Mann-Whitney U test. Categorical variables were presented as frequencies [n (%)] and evaluated using the chi-square (χ²) test.

The prediction model was validated using R version 4.4.0. RA-MCI status was defined as the dichotomous dependent variable (RA-MCI = 1; non-RA-MCI = 0). Predictors with P < 0.05 in univariate logistic regression were included in the multivariate logistic regression model to identify independent risk factors for RA-MCI.

Results

Comparison of baseline characteristics

The study ultimately enrolled 378 elderly RA patients, aged 60 to 80 years. Among them, 94 patients (24.87%) were classified into the RA-MCI group, while 284 patients (75.13%) comprised the non-RA-MCI group. Demographic and clinical data, including age, sex, COD, DAS28, OP, ESR, CRP, RF, and anti-CCP levels, were collected and analyzed. The results showed no statistically significant differences between the two groups in terms of age, sex, RF levels, or anti-CCP levels (P > 0.05). Patients in the RA-MCI group had a disease duration ranging from 1 to 39 years, with a mean duration of 16.05 ± 6.93 years. Patients in the non-RA-MCI group had a disease duration ranging from 1 to 37 years, with a mean duration of 11.32 ± 8.17 years. The disease duration was significantly longer in the RA-MCI group compared to the non-RA-MCI group. A total of 53 patients (56.38%) in the RA-MCI group had comorbid OP, significantly higher than the 108 patients (38.03%) in the non-RA-MCI group (P < 0.05). According to the DAS28 score evaluation of disease activity, the median DAS28 score in the RA-MCI group was 5.50, while the median in the non-RA-MCI group was 5.01. Furthermore, inflammatory markers were significantly elevated in the RA-MCI group compared to the non-RA-MCI group, with higher levels of both CRP and ESR (P < 0.05) (Table 1).

Table 1.

Baseline data of two groups of patients

Variables	non-RA-MCI (n = 284)	RA-MCI (n = 94)	Statistic	P
Age, [± s, year]	66.00 (64.75, 73.00)	67.00 (62.00, 72.00)	Z=-0.82	0.413
Sex, n(%)			χ²=1.76	0.185
Female	194 (68.31)	71 (75.53)
Male	90 (31.69)	23 (24.47)
COD[± s, year]	11.32 ± 8.17	16.05 ± 6.93	t=-5.48	< 0.001
OP, n(%)	108 (38.03)	53 (56.38)	χ²=9.73	0.002
DAS28, M (Q₁, Q₃)	5.01 (3.62, 6.57)	5.50 (4.98, 7.10)	Z=-3.18	0.001
CRP, [M(Q1, Q3),mg/L]	15.45 (5.78, 42.52)	38.35 (13.85, 68.38)	Z=-5.02	< 0.001
ESR[± s, mm/h]	53.61 ± 26.08	60.53 ± 26.45	t=-2.22	0.027
RF[M(Q1, Q3), mg/L]	153.00 (56.20, 448.00)	194.50 (56.20, 501.00)	Z=-1.04	0.300
Anti-CCP[M(Q1, Q3), U/ml]	100.00 (55.25, 200.00)	132.00 (50.55, 200.00)	Z=-1.09	0.276

M: Median, Q₁: 1st Quartile, Q₃: 3st Quartile, t: t-test, Z: Mann-Whitney test, χ²: Chi-square test

Disease activity was stratified in both groups based on DAS28 scores into the following categories: remission (DAS28 < 2.6), low disease activity (2.6 ≤ DAS28 ≤ 3.2), moderate disease activity (3.2 < DAS28 ≤ 5.1), and high disease activity (DAS28 > 5.1). The proportion of patients in the RA-MCI group classified as being in remission or low disease activity was significantly lower than that in the non-RA-MCI group (4.26% vs. 11.62% and 4.26% vs. 9.86%, respectively). Conversely, the RA-MCI group had significantly higher proportions of patients with moderate and high disease activity (46.81% vs. 43.31% and 44.68% vs. 35.21%, respectively). A statistically significant difference in disease activity distribution was observed between the two groups (χ²=8.41;P = 0.038) (Table 2).

Table 2.

Comparison of disease activity between two groups of patients

Variables	non-RA-MCI (n = 284)	RA-MCI (n = 94)	Statistic	P
Remission	33 (11.62)	4 (4.26)	χ²=8.41	0.038
Low disease activity	28 (9.86)	4 (4.26)
Moderate disease activity	123 (43.31)	44 (46.81)
High disease activity	100 (35.21)	42 (44.68)

χ²: Chi-square test

Regarding pharmacotherapy, conventional synthetic disease-modifying antirheumatic drugs (csDMARDs) were used by 361 patients (93.28%) in this cohort. Methotrexate (80.10%) was the most frequently prescribed csDMARD, followed by leflunomide (29.20%) and iguratimod (16.54%). Hydroxychloroquine (12.14%) and sulfasalazine (10.85%) were prescribed less frequently. Biological DMARDs (bDMARDs) were administered to 49 patients (12.66%), with tumor necrosis factor-alpha (TNF-α) inhibitors (10.08%) representing the predominant bDMARD class, and interleukin-6 (IL-6) inhibitors (1.55%) and B-cell inhibitors (1.03%) being less common. Targeted synthetic DMARDs (tsDMARDs) were used by 12 patients (3.10%). Glucocorticoids were prescribed to 184 patients (47.55%), with a median daily dose of 5 mg prednisone-equivalent. Non-steroidal anti-inflammatory drugs (NSAIDs) were used by 216 patients (55.81%), with etoricoxib (31.52%) being the most frequently administered NSAID.

Comparison of risk factors between groups

Univariate logistic regression was performed to evaluate the aforementioned risk factors. Variables were coded as follows: sex (0 = female, 1 = male), osteoporosis (OP; 0 = no, 1 = yes), with all other variables treated as continuous. The results indicated that OP prevalence, DAS28, COD, CRP, and ESR were significant risk factors for RA-MCI (Table 3).

Table 3.

Univariate logistic regression analysis of risk factors associated with RA-MCI

Variables	β	S. E	Z	P	OR (95%CI)
Age [± s, year]	0.019	0.011	1.631	0.103	1.019 (0.996 ~ 1.042)
Sex [(n, %)]	-0.359	0.272	-1.322	0.186	0.698 (0.410 ~ 1.189)
COD [± s, year]	0.072	0.015	4.704	< 0.001	1.074 (1.043 ~ 1.107)
OP (n, %)	0.745	0.241	3.088	0.002	2.107 (1.313 ~ 3.380)
DAS28, M (Q₁, Q₃)	0.234	0.071	3.282	0.001	1.263 (1.099 ~ 1.452)
CRP[M(Q1, Q3),mg/L]	0.011	0.003	4.003	< 0.001	1.011 (1.006 ~ 1.017)
ESR [± s, mm/h]	0.010	0.004	2.198	0.028	1.010 (1.001 ~ 1.019)
RF[M(Q1, Q3), mg/L]	0.000	0.000	1.833	0.067	1.000 (1.000 ~ 1.001)
Anti-CCP[M(Q1, Q3), U/ml]	0.002	0.002	1.338	0.181	1.002 (0.999 ~ 1.006)

Variables with P < 0.05 in the univariate logistic regression were included in a multivariate logistic regression model using the backward stepwise selection method. COD, OP, DAS28, and CRP were identified as independent risk factors for RA-MCI (Table 4).

Table 4.

Multivariate logistic regression analysis of independent risk factors for RA-MCI

Variables	β	S. E	Z	P	OR (95%CI)
Intercept	-4.16	0.57	-7.34	< 0.001	0.02 (0.01 ~ 0.05)
COD [± s, year]	0.07	0.02	4.07	< 0.001	1.07 (1.03 ~ 1.10)
OP [(n, %)]	0.63	0.26	2.45	0.014	1.88 (1.14 ~ 3.13)
DAS28(± s)	0.27	0.08	3.48	< 0.001	1.31 (1.13 ~ 1.53)
CRP[M(Q1, Q3),mg/L]	0.01	0.00	3.41	< 0.001	1.01 (1.01 ~ 1.02)

Construction of the RA-MCI predictive model

The predictive model was constructed using these identified risk factors, formulated as follows: Logit(P)=-4.16 + 1.07×COP (years) + 1.88×OP + 1.31×DAS28 + 1.01×CRP (mg/L). Based on the Logit(P) scores, a nomogram was developed (Fig. 2), with total scores ranging from 65 to 225, corresponding to RA-MCI risk probabilities of 0.1 to 0.9.

Fig. 2.

Nomogram for Predicting RA-MCI Risk

The discriminative performance of the nomogram was evaluated using receiver operating characteristic (ROC) analysis. The model achieved an area under the curve (AUC) of 0.750 (95% CI: 0.696–0.805), indicating moderate predictive accuracy for RA-MCI risk (Fig. 3). Model performance metrics derived from the confusion matrix were as follows: accuracy: 0.677 (95% CI: 0.628–0.724), sensitivity: 0.651 (95% CI: 0.596–0.707), specificity: 0.755 (95% CI: 0.668–0.842), positive predictive value (PPV): 0.889 (95% CI: 0.847–0.932), and negative predictive value (NPV): 0.418 (95% CI: 0.344–0.492). Calibration analysis demonstrated close agreement between predicted probabilities and observed frequencies (Hosmer-Lemeshow test, P = 0.589), confirming a good model fit (Fig. 4). Decision curve analysis (DCA) showed positive net benefit within the 10–50% threshold probability range, supporting the clinical utility of the model (Fig. 5).

Fig. 3.

ROC curve of the RA-MCI predictive model

Fig. 4.

Calibration plot of the RA-MCI predictive model

Fig. 5.

DCA of the RA-MCI predictive model

Discussion

Rheumatoid arthritis (RA), primarily affecting young to middle-aged adults, is clinically characterized by morning stiffness, joint pain, swelling, and a polyarticular symmetric pattern. As the disease progresses, it may involve vital organs such as the heart, brain, and kidneys, leading to significant functional impairment [18]. Chronic inflammation and prolonged use of immunosuppressive agents in RA patients frequently result in neurological complications [19]. Akram et al. found that demographic factors (e.g., sex, age), elevated systolic blood pressure, inflammatory markers (CRP, ESR), dyslipidemia, and comorbidities (hypertension, diabetes, atrial fibrillation, coronary artery disease) are associated with acute cerebral infarction in RA patients, while persistent systemic inflammation contributes to both physical disability and progressive cognitive decline [20]. Similarly, Chaurasia et al. reported that depression, cardiovascular disorders, and certain medications exacerbate cognitive dysfunction in this population [21]. Previous studies have demonstrated that comorbid cognitive impairment (CI) in RA patients severely compromises quality of life and increases disease burden [22]. A longitudinal study by Shin et al. further indicated that RA patients with pre-existing cardiovascular diseases face a significantly elevated risk of developing CI [23].

Mild cognitive impairment (MCI) and dementia share overlapping risk factors, with approximately 12% of MCI patients progressing to dementia annually [24]. Early prevention strategies targeting identified risk factors in MCI populations may help slow or halt disease progression [25]. A historical cohort study reported that 38–71% of RA patients develop cognitive impairment [26], while another cohort study found a prevalence of 30% [27]. The likelihood of cognitive dysfunction increases significantly in elderly RA patients [28]. Comorbid CI in RA severely compromises patients’ quality of life and exacerbates disease burden; however, early intervention can substantially improve survival and functional outcomes [22]. A systematic review of small-scale studies highlighted that RA patients exhibit a higher prevalence of cognitive dysfunction compared to healthy controls, with age, low educational attainment, disease activity, and depression identified as independent risk factors [29].

Initially, 929 RA patients were screened, of whom 405 were identified as elderly RA patients. After applying the inclusion and exclusion criteria, 378 patients were enrolled, with 94 (24.87%) diagnosed with comorbid MCI. A nationwide cross-sectional survey of cognitive function in Chinese adults aged ≥ 60 years (n = 46,011) reported an overall MCI prevalence of 15.5% [30]. The significantly higher prevalence of MCI in elderly RA patients observed in our study may be attributed to chronic systemic inflammation and long-term medication use, both of which contribute to cognitive deterioration.

Comparative analysis of risk factors between groups revealed no statistically significant differences in age, sex, RF, or anti-CCP antibody levels (P > 0.05). However, the RA-MCI group exhibited significantly longer disease duration, higher prevalence of OP, and elevated levels of DAS28, CRP, and ESR (P < 0.05). Univariate logistic regression identified disease duration, OP, DAS28, CRP, and ESR as significant risk factors for RA-MCI. Subsequent multivariate logistic regression confirmed these variables as independent predictors of RA-MCI (all P < 0.05).

In cognitive impairment disorders, advanced age is the primary risk factor, with a significant inverse correlation between age and MoCA scores [31]. In RA patients, aging may exacerbate systemic inflammation, increasing the likelihood of MCI in elderly populations [32]. However, this study focused on patients aged 60–80 years, which may explain the absence of significant age differences between groups. The DAS28, a continuous measure of RA disease activity that incorporates joint swelling and tenderness, reflects acute inflammatory burden and disease severity [33]. Our findings align with previous studies showing that elevated DAS28 levels correlate with a higher incidence of MCI, suggesting that heightened inflammatory responses accelerate cognitive dysfunction [34]. Bethany et al. similarly reported a dose-dependent relationship between DAS28 elevation and cognitive decline in RA patients [35]. Stratified analysis by DAS28 categories revealed significantly lower proportions of patients in remission and low disease activity in the RA-MCI group compared to the non-RA-MCI group, while significantly higher proportions were observed in moderate and high disease activity states. This graded association supports the hypothesis that high-grade inflammatory responses may mediate cognitive decline. Chronic inflammation in RA may drive cognitive impairment through microvascular damage, a mechanism implicated in both Alzheimer’s disease and vascular dementia [36]. Systemic inflammation and autoantibody production in RA promote vascular endothelial dysfunction and atherosclerosis, further linking inflammatory pathways to neurodegeneration [37].

Lee et al. demonstrated that RA related cognitive decline is closely associated with disease activity and inflammatory markers [38]. Among these markers, Mena-Vazquez et al. identified CRP as a critical predictor of cognitive deterioration in RA [39]. Other studies also corroborate that elevated ESR or CRP-particularly CRP-may disproportionately impact cognitive function [40]. Although RF positivity has been linked to cognitive impairment in prior research [41], no significant association was observed in our cohort. A systematic review and meta-analysis highlighted that OP increases the risk of cognitive impairment, and early OP intervention may delay its onset [42]. RA and OP share complex interactions involving immune cells, cytokines, and bone remodeling pathways, with OP being highly prevalent in elderly RA patients [43]. Our findings corroborate that OP is independently associated with cognitive dysfunction in this population, underscoring the need for integrated management of musculoskeletal and neurological health in RA.

The prediction model demonstrated moderate discrimination (AUC = 0.750) and a high positive predictive value (PPV = 0.889), indicating its efficacy in identifying RA patients at high risk of MCI. Among individuals classified as “high-risk” by the model, 88.9% were confirmed to have MCI. DCA further established significant clinical net benefit within the 10–50% threshold probability range. However, the suboptimal negative predictive value (NPV = 0.418) implies that 58.2% of patients categorized as “low-risk” were false negatives. Consequently, MCI should not be ruled out solely based on model predictions in elderly RA patients with subjective cognitive complaints or established risk factors. Comprehensive neuropsychological assessment and periodic monitoring remain essential in such cases.

The limitations of this study are as follows

First, the prediction model was developed using single-center data and was only internally validated. The absence of external validation in multicenter cohorts limits the generalizability and transportability of the model’s results. Future studies should conduct external validation in independent multicenter cohorts across diverse geographical regions and healthcare settings to assess the model’s robustness. Second, potential psychosocial confounders-particularly depressive symptoms-were not assessed. As depression is an independent risk factor for cognitive impairment, its omission may confound the true association between identified risk factors and MCI in elderly RA patients. Subsequent research should incorporate standardized neuropsychiatric inventories to comprehensively elucidate the role of psychosocial factors in RA-MCI pathogenesis. Third, several potential confounders remain unaddressed, including genetic predispositions and treatment regimen specifics (e.g., drug selection, dosage, duration). Future studies should systematically collect these data for inclusion as covariates or for model optimization to better control confounding effects. Fourth, the substantially smaller sample size in the RA-MCI group may compromise the stability of multivariate analyses. Future investigations should prioritize the recruitment of RA-MCI patients to balance group sizes and employ machine learning algorithms for exploratory analyses of imbalanced data. Finally, the suboptimal negative predictive value (NPV = 0.418) increases false-negative rates. Subsequent model refinements should integrate additional clinical parameters, validated biomarkers, and advanced modeling techniques to improve the accuracy of risk stratification for MCI in elderly RA populations.

Conclusions

In summary, research on CI comorbidities in RA remains limited. Cognitive dysfunction in elderly RA patients significantly complicates chronic disease management, requiring tailored treatment optimization and rigorous outcome monitoring. COD, OP, elevated DAS28, and high CRP levels are strong predictors of MCI in this population. The early predictive model developed in this study demonstrated high accuracy in screening for cognitive deficits, providing a clinically actionable tool to guide early interventions. By identifying key risk factors and constructing risk stratification models, clinicians can proactively mitigate cognitive decline in elderly RA patients, thereby delaying the onset of dementia. While the nomogram developed in this study demonstrated favorable discrimination and clinical utility in internal validation, its primary application remains in risk stratification and initial screening. Future efforts should focus on enhancing model sensitivity and negative predictive value through algorithmic optimization, conducting multicenter external validation across diverse populations to strengthen generalizability, and implementing model calibration refinements to improve overall predictive capability.

Abbreviations

RA

Rheumatoid arthritis

CI

cognitive impairment

MCI

Mild cognitive impairment

MoCA

Montreal Cognitive Assessment

ROC

receiver operating characteristic

DCA

decision curve analysis

COD

Duration of Diseased

DAS28

Disease Activity Score-28

CRP

C-reactive protein

OP

Osteoporosis

AD

Alzheimer’s disease

ESR

Erythrocyte sedimentation rate

RF

rheumatoid factor

anti-CCP

Anti-cyclic citrullinated peptide; standard deviation (SD)

Author contributions

JY drafted the paper. HG and LXZ made substantive revisions. YKD and JL collected data. PC, YYG, YN conducted statistical analysis. All authors reviewed the manuscript.

Funding

Shaanxi Provincial Key Research and Development Program (2019SF-192, 2022SF-266, 2025SF-YBXM-052); Traditional Chinese Medicine Technology Innovation and Capacity Expansion Plan (TZKN-CXPT-04); Xi’an Innovation Capability Strengthening Program (24YXYJ0076,2024YXYJ0107); Research Project of Xi’an Municipal Health Commission(2025yb26); In-house Fund of the Fifth Hospital of Xi’an (2023lc07, 2024lc03, 2024lc04).

Data availability

Data generated or analyzed during this study are included in this published article and its supplementary files. The de-identified datasets are available from the corresponding author upon reasonable request, subject to approval by the Ethics Committee of Xi’an Fifth Hospital.

Declarations

Ethics approval and consent to participate

This study adhered to the principles of the Declaration of Helsinki and was approved by the Medical Ethics Committee of Xi’an Fifth Hospital (Approval No.: [2023] Ethics Review 55). The trial was registered with the Chinese Clinical Trial Registry (Registration No.: ChiCTR2300077337, Registration Date: 2023-11-01). Written informed consent was obtained from all individual participants included in the study.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.