Objective TCM syndrome differentiation in rheumatoid arthritis patients with cerebral infarction: The auxiliary diagnostic value of autoantibodies (anti-CCP IgG, AKA, and GPI) and hemoglobin

Abstract

The co-occurrence of rheumatoid arthritis (RA) and cerebral infarction is highly prevalent in clinical practice. While integrated traditional Chinese medicine (TCM) and Western medicine offers unique advantages in treatment, the objective indicators associated with TCM syndromes in this specific patient population remain unclear, hindering precise syndrome differentiation. This study, utilizing a screened cohort from 920 hospitalized patients with RA and cerebral infarction, aimed to address this research gap. Based on strict inclusion, exclusion, and elimination criteria, 142 patients were selected from the initial 920. The distribution of TCM syndromes and their associated influencing factors were analyzed using the Kolmogorov–Smirnov test, Brown–Forsythe ANOVA, the chi-square test, as well as both binary and multinomial logistic regression (employed complementarily to overcome the sample size limitations of less common syndromes). First, among the quantitative indicators, only hemoglobin (Hb) level showed significant differences between groups. The Hb level in the wind–cold obstruction syndrome was significantly higher than that in the dampness–heat obstruction syndrome (P = .007) and the phlegm–stasis obstruction syndrome (P = .029). Second, glucose-6-phosphate isomerase, anti-keratin antibody, and anti-cyclic citrullinated peptide IgG were significantly associated with TCM syndromes (P < .05). Specifically, they were identified as independent risk factors for dampness–heat obstruction syndrome (OR = 2.611, 2.218), while also serving as independent protective factors for liver–kidney deficiency syndrome relative to phlegm–stasis obstruction syndrome (OR = 0.294, 0.350). Within the studied population from East China, glucose-6-phosphate isomerase, anti-keratin antibody, anti-cyclic citrullinated peptide IgG, and Hb show associations with TCM syndrome differentiation in patients with RA complicated by cerebral infarction, particularly for the dominant damp–heat obstruction syndrome. The rigorous screening process enhances the reliability of the conclusions for the major syndromes studied, providing a preliminary evidence-based foundation for objective syndrome differentiation in this specific clinical context. Further multi-center studies with larger samples, especially of rare syndrome types, are needed to validate and generalize these findings.

1. Introduction

Cerebral infarction,^[1] as the most common type of cerebrovascular disease, accounts for approximately 70% of all acute cerebrovascular events; among hospitalized patients with cerebrovascular diseases in China, ischemic stroke constitutes up to 83%.^[2] The China Stroke Prevention and Treatment Report 2020 indicates that the incidence of ischemic stroke in China is 145 per 100,000 people, with the number of patients aged 40 and above reaching 17.04 million. Its characteristics of high incidence, high disability rate, and high mortality pose a serious threat to national health.^[3,4] Rheumatoid arthritis (RA) is a chronic autoimmune disease characterized by erosive, symmetric polyarthritis. It not only leads to joint deformity and functional impairment but also affects the overall health of patients through systemic inflammatory responses.^[5] The pathological process of RA burdens not only the physical health of patients but also significantly impacts their psychological and social functioning.^[6,7] Recent studies have confirmed that RA patients have a significantly increased risk of cardiovascular and cerebrovascular events, with their risk of cerebral infarction markedly higher than that of the general population.^[8,9] This association primarily stems from chronic inflammation-mediated vascular injury and potential effects of therapeutic drugs.^[10]

When RA (known as “Wang Bi” in Traditional Chinese Medicine, TCM) coexists with cerebral infarction (categorized as “ischemic stroke” in TCM), a special state of pathological overlap is formed. The pathogenesis of Wang Bi is based on liver–kidney deficiency and qi–blood deficiency,^[11–13] with invasion of wind–cold–dampness evils leading to phlegm–stasis coagulation and joint deformation; meanwhile, the core pathogenesis of ischemic stroke involves visceral dysfunction, disruption of qi and blood, and phlegm–stasis intermingling obstructing the brain collaterals.^[14,15] The 2 diseases share significant commonalities in pathological progression: long-term use of immunosuppressants by RA patients may induce a hypercoagulable state, increasing stroke risk; whereas limited limb mobility in cerebral infarction patients can aggravate joint stiffness and other Bi syndrome manifestations, forming a “pathological interaction” cycle.^[16–18] This dual pathological state significantly increases clinical treatment difficulty: the needs for joint functional rehabilitation and neurological function training often conflict. Patients not only face physical problems such as pain and dyskinesia but also endure multiple burdens including impaired mental health and reduced quality of life.^[19] When patients suffer from both diseases simultaneously, their quality of life further declines, placing heavy pressure on families and society.^[20,21] Therefore, identifying pathogenic factors correlated with both diseases is particularly important to alleviate patients’ suffering and improve their quality of life.^[22]

Western medical research has revealed a deeper intrinsic link between the 2: cerebral infarction is a syndrome group of diseases resulting from lack of blood supply causing local brain tissue necrosis, with symptoms depending on the location and area of infarction.^[23] Treatment of cerebral infarction primarily involves thrombolysis, anticoagulation, and blood pressure regulation to alleviate symptoms and prevent further deterioration.^[24] The chronic inflammatory response in RA can form a “inflammation–thrombosis” vicious cycle with cerebral infarction through mechanisms such as vascular endothelial injury and promotion of atherosclerosis.^[25] The 2 conditions share cardiovascular risk factors, and the persistent inflammatory burden in RA patients makes them more susceptible to stroke events.^[26] RA can be classified into different clinical types based on disease severity and joint involvement, such as early mild type, polyarticular type, oligoarticular type, and severe type. Current Western treatments emphasize early intervention and personalized regimens, with overlapping consensus on inflammation control, antithrombotic strategies, and vascular protection^[27]; meanwhile, TCM’s understanding of both diseases focuses on treatment based on syndrome differentiation, classifying stroke into various syndrome types, and has accumulated rich theoretical and practical experience in treating Wang Bi.^[28] However, research on the distribution patterns of TCM syndromes in RA complicated with cerebral infarction and their correlation with objective detection indicators remains relatively scarce.

This study is based on real-world clinical data from the First Affiliated Hospital of Anhui University of Chinese Medicine. From 920 patients hospitalized due to RA complicated with cerebral infarction, 142 patients were strictly selected according to the inclusion, exclusion, and elimination criteria detailed below. A systematic analysis of TCM syndrome characteristics was conducted, and statistical methods were used to explore the correlations between different syndrome types and relevant laboratory indicators (including measurement data and count data). The study aims to identify objective indicators closely related to disease progression, provide a basis for establishing a precision diagnosis and treatment system integrating Chinese and Western medicine, thereby optimizing early diagnosis processes, developing personalized treatment plans, and offering theoretical support for disease prevention strategies.

2. Methods

2.1. Study subjects

This study conducted a retrospective analysis of 142 patients admitted to the Department of Rheumatology, the First Affiliated Hospital of Anhui University of Chinese Medicine between March 2021 and May 2024, all diagnosed with RA complicated by cerebral infarction. The cohort included 26 male patients (mean age: 68.3 years) and 116 female patients (mean age: 63.6 years), with ages extracted from the front pages of medical records. The anonymized baseline data and analyzed variables for all patients are available as supplementary material (see Data S1, Supplemental Digital Content, https://links.lww.com/MD/R346) (notably, binary variables use 0/1 coding: 0 = negative/normal, 1 = positive/abnormal).

This study is a retrospective cohort analysis. The subjects were patients with RA complicated by cerebral infarction who were admitted to the Department of Rheumatology of the First Affiliated Hospital of Anhui University of Chinese Medicine between March 2021 and May 2024. A total of 920 cases meeting the preliminary screening criteria were retrieved from the hospital’s electronic medical record (EMR) system. After strict screening based on the inclusion, exclusion, and elimination criteria detailed in Sections 2.3 to 2.5, 142 patients were ultimately included for statistical analysis (all patient baseline data can be viewed in Data S1, Supplemental Digital Content, https://links.lww.com/MD/R346).

The baseline characteristics of the included patients are as follows: 26 males (18.31%), age range 52 to 81 years, mean age (68.3 ± 7.5) years; 116 females (81.69%), age range 48 to 79 years, mean age (63.6 ± 6.8) years. Patient age data were extracted from the “actual age at admission” field in the inpatient medical record homepage. Baseline information such as gender, time of onset, and medical history were verified through the EMR system and paper records to ensure data authenticity.

This work has been reviewed and approved by the Ethics Committee of the First Affiliated Hospital of Anhui University of Chinese Medicine (Ethics Approval No.: 2023AH-52). The research process strictly adhered to the principles of the Declaration of Helsinki. Patient privacy information (such as name, ID number, hospitalization number, etc) was de-identified, with only relevant codes retained for data traceability.

2.2. Diagnostic and TCM syndrome classification criteria

2.2.1. Western medical diagnostic criteria

RA diagnosis: refer to the “2010 American College of Rheumatology (ACR)/European League Against Rheumatism (EULAR) Classification Criteria for Rheumatoid Arthritis”.^[29] This criteria includes 4 dimensions: joint involvement (e.g., number and type of affected joints); serological markers (positivity and titer of anti-cyclic citrullinated peptide antibody [anti-CCP IgG] and rheumatoid factor); duration of synovitis (≥6 weeks); acute phase response indicators (elevated C-reactive protein and erythrocyte sedimentation rate). A total score of ≥6 points across these 4 dimensions confirms RA diagnosis.

Cerebral infarction diagnosis: conforms to the “Chinese Guidelines for Diagnosis and Treatment of Acute Ischemic Stroke 2018”.^[30] The diagnosis requires meeting all of the following: acute onset with neurological deficit symptoms (e.g., limb weakness, speech impairment, dysphagia); cranial computed tomography or magnetic resonance imaging excluding cerebral hemorrhage and confirming the presence of an ischemic brain lesion consistent with symptoms; time since onset ≤ 2 weeks (or for chronic-phase patients, clear history of acute onset and follow-up imaging evidence).

2.2.2. TCM syndrome differentiation criteria

TCM syndrome differentiation was centered on the syndrome elements of “Wang Bi” (corresponding to RA) and “ischemic stroke” (corresponding to cerebral infarction), with reference to 2 authoritative standards: “National Standard of the People’s Republic of China – Clinical Terminology of Traditional Chinese Medicine (Syndromes Part)”,^[31] which specifies the naming conventions and core elements (e.g., disease location, nature) for Bi syndrome (including Wang Bi) and stroke syndromes; “Traditional Chinese Medicine Terminology” promulgated by the Terminology Committee of Traditional Chinese Medicine,^[32] which provides standardized definitions and clinical manifestations for syndromes such as “cold–dampness obstruction syndrome” and “phlegm–stasis obstruction syndrome.” Typing integrated joint symptoms of RA and neurological symptoms of cerebral infarction to form a “disease–syndrome” combined differentiation system.

2.3. Inclusion criteria

1. Western medical diagnosis meets the aforementioned “2010 ACR/EULAR RA Classification Criteria” and “Chinese Guidelines for Diagnosis and Treatment of Acute Ischemic Stroke 2018”, with clear coexistence of RA and cerebral infarction (RA diagnosis may precede or follow cerebral infarction, but clinical evidence of both diseases must be present at admission).

2. Age ≥ 18 years (due to significant differences in pathogenesis of RA and cerebral infarction between minors and adults, only adult cases were included), Gender is not limited (unrestricted).

3. No severe cognitive impairment (e.g., Alzheimer disease, vascular dementia) or Communication barrier (e.g., severe aphasia), able to cooperate with clinical examinations (e.g., joint range of motion assessment, neurological function scoring) and objective indicator testing (e.g., blood tests, imaging studies).

4. Complete medical records during hospitalization and within 1 month after discharge, with no missing key information (e.g., 4 diagnostic information [inspection, listening/smelling, inquiry, palpation], test results, treatment plans, follow-up records).

5. The patient or their legal guardian has signed the “Informed Consent Form for Use of Medical Record Data in Research” (For retrospective cases where the patient could not be contacted, an “informed consent waiver” procedure was applied after approval by the Ethics Committee).

2.4. Exclusion criteria

1. Meeting diagnostic criteria for only RA or cerebral infarction alone (e.g., RA without history or imaging evidence of cerebral infarction, or cerebral infarction without history or serological evidence of RA).

2. RA diagnosis being “undifferentiated arthritis,” “psoriatic arthritis,” or other rheumatic diseases, or cerebral infarction being “hemorrhagic transformation type,” “cerebral embolism (e.g., cardiogenic embolism)” or other non-primary ischemic stroke.

3. Comorbid severe mental disorders (e.g., acute phase schizophrenia, manic episode of bipolar disorder) or severe somatic diseases (e.g., advanced malignant tumors, acute liver or kidney failure, active acute infections such as pneumonia, urinary tract infection, or sepsis requiring intravenous antibiotics), which may render the patient unable to cooperate with research-related examinations or could significantly and transiently affect the stability and interpretability of core inflammatory indicators (e.g., high-sensitivity C-reactive protein [hs-CRP], white blood cell count [WBC]).

4. Missing ≥ 2 items of key information in medical records (e.g., missing any 2 or more of RA inflammation markers, cerebral infarction imaging results, TCM tongue and pulse records), making syndrome determination or indicator analysis impossible.

5. Receiving special treatments such as immunoadsorption or biological agents (e.g., TNF-α antagonists) within 3 months prior to admission, which may affect the detection results of objective indicators.

2.5. Elimination criteria

1. After inclusion, data could not be completely recorded due to patient self-discharge (without completing the treatment course), loss to follow-up, or discovery of missing core data (e.g., post-treatment review indicators, syndrome reconfirmation results) upon supplementary review of medical records.

2. Occurrence of acute conditions during the research process (e.g., acute myocardial infarction, pulmonary embolism, severe pneumonia) requiring termination of the original disease treatment plan, or patient death (making subsequent follow-up data unavailable).

3. After independent review by 2 TCM physicians, serious disagreement in syndrome determination was found (and consensus could not be reached through arbitration by a third chief physician), or diagnostic error was discovered (e.g., misdiagnosis of “osteoarthritis” as RA).

2.6. Syndrome differentiation method

2.6.1. Differentiation implementation process

A syndrome differentiation expert group consisting of 3 TCM physicians (all with the title of associate chief physician or above, and ≥10 years of clinical experience in rheumatology or encephalopathy departments) was established, adopting a “double-blind independent differentiation + cross-review” model:

Data extraction: TCM 4 diagnostic data (inspection, listening/smelling, inquiry, palpation) at patient admission were extracted from the EMR, including symptoms (e.g., nature of joint pain, presence of chills/fever, dizziness/headache, limb numbness), signs (e.g., degree of joint swelling, muscle strength grading), tongue examination (tongue body, tongue coating color and thickness), and pulse examination (pulse type such as wiry, slippery, thin, deep, etc).

Independent differentiation: 2 physicians independently determined the patient’s syndrome type with reference to the “National Standard of the People’s Republic of China – Clinical Terminology of Traditional Chinese Medicine Part 2: Syndromes” (GB/T 16751.2-2021), the “Traditional Chinese Medicine Terminology”, and “Diagnostic Elements of TCM Syndromes in RA Complicated with Cerebral Infarction” (see Table 1, including core symptoms, accompanying symptoms, tongue, and pulse criteria for each syndrome type).

Table 1.

Diagnostic elements of each syndrome type.

Syndrome name	Core symptoms	Tongue signs	Pulse signs
Cold–dampness obstruction syndrome	Cold pain and heaviness in limbs/joints, fixed pain location, worsens with cold and alleviates with warmth, limited mobility, local swelling or numbness	Light red tongue with white, greasy coating	Taut-tight or soft-slow pulse
Damp–heat obstruction syndrome	Red, swollen, hot, and painful joints, burning sensation, refusal to touch, limited mobility, accompanied by fever, thirst, yellow urine	Red tongue with yellow, greasy coating	Slippery-rapid pulse
Phlegm–stasis obstruction syndrome	Stabbing pain in joints with fixed location, swelling/deformation, limited mobility, subcutaneous nodules or , worse at night	Dark purple tongue with white/yellow greasy coating	Taut-stringy or slippery pulse
Liver–kidney deficiency syndrome	Dull joint pain, soreness in lower back and knees, dizziness, fatigue, worsens with exertion;	Light red tongue (pale-swollen for yang deficiency, red with little coating for yin deficiency)	Deep-thin pulse (deep-slow for yang deficiency, thin-rapid for yin deficiency)
Spleen–kidney deficiency syndrome	Joint pain, fatigue, poor appetite, loose stools, edema, shortness of breath, fear of cold	Pale-swollen tongue with white coating	Deep-weak pulse
Wind–cold obstruction of collaterals syndrome	Wandering joint pain (wind dominance) or severe fixed pain (cold dominance), worsens with cold, accompanied by aversion to wind and fever	Light red tongue with thin white coating	Floating-tight (wind dominance) or taut-tight (cold dominance) pulse

Review and arbitration: if the syndrome determinations by the 2 physicians were consistent, this was taken as the final syndrome type. In case of disagreement, a third physician performed a blinded review, and the majority opinion was taken as the final syndrome type (if all 3 opinions differed, the case was handled according to “Elimination Criterion 3”).

2.6.2. Criteria for syndrome differentiation

2.6.2.1. Basis for syndrome selection

This study selected 6 syndrome types: “cold–dampness obstruction syndrome, dampness–heat obstruction syndrome, phlegm–stasis obstruction syndrome, liver–kidney deficiency syndrome, spleen–kidney deficiency syndrome, and wind–cold obstruction of collaterals syndrome.” The selection was primarily based on the following 4 aspects:

1. Pathomechanism relevance: the core pathogenesis of RA (Wang Bi) is “obstruction of meridians and deficiency of viscera,” while that of cerebral infarction (ischemic stroke) is “disorder of qi and blood and obstruction of brain collaterals.” Both share key pathological links of “obstruction” (caused by cold, heat, phlegm, stasis, or wind pathogens) and “deficiency” (deficiency of viscera such as liver, spleen, and kidney). These 6 syndromes comprehensively cover the 2 major pathogenetic types: “excess evil” (cold–dampness, dampness–heat, phlegm–stasis, wind–cold) and “root deficiency” (liver–kidney deficiency, spleen–kidney deficiency), aligning with the “disease–syndrome” combined differentiation logic.

2. Clinical prevalence: a preliminary retrospective analysis of 316 cases of RA complicated with cerebral infarction in our hospital from 2018 to 2020 showed that the above 6 syndromes accounted for 96.4% of cases (with phlegm–stasis obstruction syndrome and liver–kidney deficiency syndrome being the most frequent, at 28.5% and 23.1%, respectively). The remaining syndromes (such as qi–blood deficiency syndrome and phlegm–heat fu-organ excess syndrome) accounted for <4%. Due to insufficient sample size for statistical analysis, these were not included.

3. Academic standardization: all 6 syndromes are clearly included in the “National Standard of the People’s Republic of China – Clinical Terminology of Traditional Chinese Medicine Part 2: Syndromes”, the “Traditional Chinese Medicine Terminology”, the “Guidelines for the Diagnosis and Treatment of Rheumatoid Arthritis in Traditional Chinese Medicine (2021)”, and the “Guidelines for the Treatment of Ischemic Stroke with Chinese Patent Medicines in China (2022)”. The diagnostic elements (symptoms, tongue, and pulse) are standardized and unambiguous, facilitating consistent differentiation among physicians.

4. Clinical practicality: the treatment principles corresponding to the 6 syndromes (e.g., warming meridians and dispelling cold to unblock collaterals for cold–dampness obstruction syndrome; resolving phlegm and activating blood to unblock collaterals for phlegm–stasis obstruction syndrome) are commonly used in clinical practice. Clarifying their association with objective indicators can directly provide references for subsequent individualized treatment, aligning with the research objective of “scientific research serving clinical practice.”

2.6.2.2. Core diagnostic elements of each syndrome type

The core diagnostic elements for each syndrome were strictly based on Table 1, as detailed below.

Cold–dampness obstruction syndrome: core symptoms include cold pain and heaviness in limb joints, fixed pain, worsening pain with cold and alleviation with heat, limited joint extension and flexion, local swelling or skin numbness; tongue: pale red with white greasy coating; pulse: wiry-tight or soft-moderate.

Dampness–heat obstruction syndrome: core symptoms include red, swollen, hot, and painful joints with burning sensation and tenderness upon pressure, limited movement, accompanied by fever, thirst, and dark yellow urine; tongue: red with yellow greasy coating; pulse: slippery-rapid.

Phlegm–stasis obstruction syndrome: core symptoms include fixed stabbing pain in joints, swelling and deformity, limited flexion and extension, subcutaneous nodules or ecchymosis, worse pain at night; tongue: dark purple or with ecchymosis, with white or yellow greasy coating; pulse: wiry-choppy or slippery.

Liver–kidney deficiency syndrome: core symptoms include dull pain in joints, soreness and weakness in lower back and knees, dizziness and fatigue, worsening with exertion; yin deficiency may present with 5-center heat, yang deficiency with cold limbs and aversion to cold; tongue: pale red (enlarged and pale for yang deficiency, red with scant coating for yin deficiency); pulse: deep-thin (deep-slow for yang deficiency, thin-rapid for yin deficiency).

Spleen–kidney deficiency syndrome: core symptoms include joint pain, fatigue, poor appetite, loose stools, edema, shortness of breath, and aversion to cold; tongue: pale and enlarged with white coating; pulse: deep-weak.

Wind–cold obstruction of collaterals syndrome: core symptoms include migratory joint pain (wind predominance) or severe fixed pain (cold predominance), worsening with cold, accompanied by aversion to wind and fever; tongue: pale red with thin white coating; pulse: floating-tight (wind predominance) or wiry-tight (cold predominance).

2.6.3. Quality control in syndrome differentiation

Quality control was implemented using the Design, Measurement, and Evaluation method for clinical research.

Before differentiation, the expert group received training to unify the judgment criteria for diagnostic elements of each syndrome (e.g., “cold pain in joints” was defined as “pain accompanied by an aversion to cold sensation, with local skin temperature of the joint lower than normal”).

A random sample of 20% of the cases (28 cases in total) was selected for “test–retest reliability” assessment, and the Kappa coefficient was calculated (Kappa ≥ 0.75 was considered indicative of good consistency in syndrome differentiation).

Cases with disagreements during the differentiation process were recorded (e.g., “yellow greasy coating vs yellow dry coating”) and included in the research summary for discussion to ensure the objectivity and reliability of differentiation results.

2.7. Influencing factors

A total of 142 patients with RA complicated by cerebral infarction were included in this study. After strict exclusion of cases with missing data, the objective indicators ultimately included for analysis comprised the following clinical data and laboratory parameters:

Age: obtained from the patient’s hospitalization record homepage.

TCM syndrome type: classified according to the criteria outlined in Section 2.6.

Biochemical indicators: including hs-CRP, glucose (Glu), high-density lipoprotein cholesterol (HDL-C), low-density lipoprotein cholesterol (LDL-C), triglycerides (TG), and total cholesterol (TC).

RA-related antibody indicators: including glucose-6-phosphate isomerase (GPI), anti-keratin antibody (AKA), and anti-CCP IgG.

Anti-neutrophil cytoplasmic antibody (ANCA) profile: including cytoplasmic anti-neutrophil cytoplasmic antibody (c-ANCA), anti-glomerular basement membrane antibody, anti-myeloperoxidase antibody (anti-MPO), formalin-resistant ANCA, formalin-sensitive P-ANCA, and anti-proteinase 3 antibody.

Complete blood count indicators: including WBC, red blood cell count (RBC), hemoglobin (Hb), absolute neutrophil count (ANC), and platelet count (PLT). Venous blood samples for the aforementioned laboratory indicators were collected from all patients under fasting conditions on the morning following admission. This standardized protocol aimed to minimize interference from diet and diurnal rhythm on test results.

Autoantibody profile: including anti-centromere protein B antibody, centromere pattern, nuclear dots pattern, anti-double stranded DNA antibody, cytoplasmic fibrous pattern, cytoplasmic granular pattern, anti-histone antibody, homogeneous nuclear pattern (HNP), anti-Jo-1 antibody, nuclear membrane pattern, anti-U1 ribonucleoprotein/Smith antibody, nucleolar pattern, anti-nucleosome antibody (ANA) [note: often ANA refers to anti-nuclear antibody screening, specific confirmatory tests like anti-nucleosome are specified], speckled nuclear pattern, anti-Ro-52 antibody, anti-ribosomal P protein antibody, anti-Scl-70 antibody, anti-Smith antibody (anti-Sm), anti-Sjögren syndrome-related antigen A antibody (anti-SSA), anti-Sjögren syndrome-related antigen B antibody.

Venous blood samples for the aforementioned laboratory indicators were collected from all patients under fasting conditions in the morning after admission. All patients underwent cranial magnetic resonance imaging or computed tomography examination.

In terms of data types, age, hs-CRP, Glu, HDL-C, LDL-C, TG, TC, WBC, RBC, Hb, ANC, and PLT were treated as measurement data; all other indicators were treated as categorical data.

This study also retrospectively collected other clinical information that could act as confounding factors, including: RA disease duration (from diagnosis), history of RA-related medications prior to admission (e.g., use of methotrexate, leflunomide, glucocorticoids, biologics), cerebral infarction severity (estimated NIHSS score range based on medical records), and comorbidities (history and treatment of hypertension, diabetes, hyperlipidemia). In preliminary analyses, we recognized that directly incorporating all these variables into the multivariate models could lead to overfitting due to complex interactions and incomplete recording for some variables. Therefore, the primary regression analyses in this study focused on the direct associations of demographic, core laboratory, and antibody indicators with syndrome types, while treating the aforementioned clinical factors as important variables for stratification or sensitivity analyses (see Section 3.4.2).

This study focuses on key indicators such as hs-CRP, GPI, AKA, HNP, HDL-C, LDL-C, and Glu, which collectively form a core network linking multiple pathological processes including immune-inflammatory responses, vascular endothelial injury, lipid metabolism disorders, and neural damage. The dynamic changes in these indicators not only reflect the disease progression and pathological mechanisms of RA complicated with cerebral infarction but also provide a crucial basis for the objective and quantitative evaluation of TCM syndromes. Building upon this, this study will further analyze the associations between different TCM syndrome types and the objective indicators mentioned above, aiming to reveal/clarify their intrinsic patterns and provide more targeted objective evidence for integrated TCM and Western medicine diagnosis and treatment.

2.8. Statistical methods

2.8.1. Data management and quality control

A standardized database was first established. Data were entered using a “double data entry + cross-verification” model: 2 trained researchers independently entered the baseline data (gender, age), TCM syndrome types, and laboratory indicators (measurement/categorical data) of the 142 patients into EpiData 3.1 software (The EpiData Association, Odense, Denmark). After entry, the software’s built-in logic check functions (e.g., range checks: age > 120 years flagged as an outlier, inflammation indicators < 0 flagged as missing) were used to identify erroneous data. A 3rd researcher then manually verified the data against the original medical records to ensure a data entry accuracy rate of ≥99%. For outliers identified during verification (e.g., serum CRP > 100 mg/L), clinical records were consulted to determine if they represented true values (e.g., patient with concurrent acute infection). True values were retained and annotated, while erroneous data were corrected and reentered.

Missingness for all variables was checked during the data cleaning phase. The core variables included in this study (required for syndrome differentiation and as primary outcome measures, such as data for the 4 diagnostic methods and key laboratory results like hs-CRP and antibody profiles) were mandated to be complete per the inclusion/exclusion criteria. Consequently, the missing rate for these core variables in the final cohort of 142 patients was 0%. For a very small number of non-core variables or items (e.g., certain autoantibodies not routinely tested, where absence was treated as “not detected”), complete-case analysis was used for analyses involving that specific variable. Given the high completeness of the core analytical variables, we considered the risk of bias introduced by complete-case analysis to be low. The effective sample size for each primary analysis variable is indicated in the corresponding results tables.

2.8.2. Analysis of measurement data

All measurement data (e.g., age, hs-CRP) were 1st tested for normal distribution and homogeneity of variance.

Normality test: the Shapiro–Wilk test was used (applicable for sample size n < 500; this study n = 142). A P-value > .05 indicated that the data followed a normal distribution.

Homogeneity of variance test: Levene test was used. A P-value > .05 indicated homogeneity of variance across groups.

The appropriate analysis method was selected based on the test results.

Measurement data satisfying “random sample + normal distribution + homogeneity of variance” were expressed as mean ± standard deviation ($\bar{x}$ ± s). Comparisons between groups were performed using 1-way analysis of variance (1-way ANOVA).

Measurement data satisfying random sample and normal distribution but with heterogeneity of variance were analyzed using Brown–Forsythe ANOVA (which is more robust to heteroscedastic data than traditional ANOVA).

Measurement data not following a normal distribution were expressed as median (interquartile range) [M (Q1, Q3)]. Comparisons between groups were performed using the non-parametric Kruskal–Wallis H test.

If the overall difference was statistically significant (P < .05) in the above analyses, post hoc multiple comparisons were conducted: LSD test was used when variances were homogeneous (suitable for pairwise comparisons with balanced sample sizes); Dunnett T3 test was used when variances were heterogeneous; the Dunn test was used for non-normally distributed data (with the significance level corrected to α = 0.05/number of comparisons to avoid inflation of type I error).

2.8.3. Analysis of categorical data

Categorical data (e.g., gender) were expressed as number (percentage) [n (%)]. Comparisons between groups were performed using Pearson chi-square test (χ² test), strictly adhering to the following application conditions.

When the theoretical frequency (T) in all cells was ≥5 and the total sample size (n) was ≥40, the standard Pearson χ² test was used.

When any cell had 1 ≤ T < 5 and n ≥ 40, the Yates corrected Pearson χ² test was used.

When any cell had T < 1 or n < 40, Fisher exact test (2-sided) was used to avoid reduced test power due to small theoretical frequencies.

If analyzing the association between TCM syndrome type and a categorical indicator (e.g., the relationship between phlegm–stasis obstruction syndrome and history of hypertension), the contingency coefficient (C coefficient) or Phi coefficient could be further calculated to determine the association strength (|coefficient| between 0.1 and <0.3 indicating weak association, between 0.3 and <0.5 indicating moderate association, and ≥0.5 indicating strong association).

2.8.4. Univariate and multivariate analysis

1. Univariate analysis: with “TCM syndrome type” as the dependent variable (the 6 syndrome types were set as categorical variables, e.g., cold–dampness obstruction syndrome = 1, dampness–heat obstruction syndrome = 2,..., wind–cold obstruction of collaterals syndrome = 6) and various objective indicators (e.g., age, gender, inflammation markers, underlying diseases) as independent variables, the analysis methods for measurement/categorical data described above were used to screen for statistically significant variables (variables with P < .05 were included in the multivariate analysis) and exclude non-associated variables (P ≥ .05) to reduce model interference.

2. Regression model construction strategy: multinomial logistic regression analysis: used as the core analysis model. Since the TCM syndrome type is a multi-category variable (6 types), a multinomial logistic regression model was used to holistically explore the independent association of various indicators with the syndrome types. The most common syndrome type (preliminary analysis indicated phlegm–stasis obstruction syndrome had the highest prevalence, hence set as the reference group) was used as the reference. Independent variables were handled as follows: categorical independent variables (e.g., gender: male = 1, female = 0; hypertension: yes = 1, no = 0) were entered directly; continuous independent variables (e.g., age, ESR) were 1st standardized (Z-score = (x − μ)/σ) to eliminate scale effects. Variables were selected using the Forward Stepwise (Likelihood Ratio, LR) method, with entry criteria α = 0.05 and removal criteria α = 0.10. Model reliability was verified using the Hosmer–Lemeshow goodness-of-fit test (P > .05 indicates good model fit) and Variance Inflation Factor (VIF < 10 indicates no significant multicollinearity). Due to the highly skewed gender ratio in our study (81.69% female), which is an inherent epidemiological feature of RA, we did not force gender into the primary models as an adjustment variable to avoid over-adjustment that might mask true biological associations. However, we recognize gender as a potential confounder. Therefore, in a sensitivity analysis (detailed in Section 3.4.2), we included gender in the models for verification and found that the main association results remained stable, suggesting that the core findings (e.g., the association of GPI and AKA with damp–heat obstruction syndrome) are independent of the gender distribution. Formal gender-stratified analysis in future larger-scale studies would be valuable.

Binary logistic regression analysis: used as a supplementary analysis for in-depth exploration of specific syndrome types of particular clinical interest (e.g., dampness–heat obstruction syndrome). The target syndrome was set as the case group (coded = 1), and all other syndromes were combined into the control group (coded = 0) to construct a binary logistic regression model. The assignment of independent variables, selection method (Forward Stepwise LR), and model verification methods (Hosmer–Lemeshow test, VIF check) remained consistent with the multinomial model. To control the risk of multiple comparisons, the Bonferroni correction was applied to the significance level (corrected α = 0.05/ number of syndrome types analyzed). This analysis aimed to precisely identify factors independently associated with a specific syndrome type, supplementing potential characteristics of particular syndromes that might be obscured in the overall comparison of the multinomial model, analogous to the strategy of separate analysis for dampness–heat amassment syndrome in research on TCM syndromes of liver cancer.

2.8.5. Statistical software and significance level

All statistical analyses were performed using SPSS software (version 26.0; IBM Corporation, Armonk), and graphing was conducted using GraphPad Prism (version 9.0; GraphPad Software, Inc., La Jolla). The significance level was uniformly set at α = 0.05 (2-tailed test). A P-value < .05 was considered statistically significant for differences or associations. The technical roadmap of this study is shown in Figure 1.

Figure 1.

Technical roadmap: rheumatoid arthritis complicated with cerebral infarction.

3. Results

3.1. Distribution of patient syndrome types

A total of 142 patients with RA complicated by cerebral infarction were included in this study. Based on the predefined TCM syndrome differentiation criteria (referencing the National Standard of the People’s Republic of China – Clinical Terminology of Traditional Chinese Medicine Part 2: Syndromes and the Traditional Chinese Medicine Terminology), the patients were classified into 6 syndrome types. The distribution frequency, composition ratio, and cumulative composition ratio of each syndrome type are detailed in Table 2, and the visual distribution is presented in Figure 2.

Table 2.

Statistical table of certificate distribution.

	Frequency	Percentage	Cumulative percentage
Cold–dampness obstruction syndrome	2	1.41%	1.41%
Dampness–heat obstruction syndrome	82	57.75%	59.15%
Liver–kidney deficiency syndrome	27	19.01%	78.17%
Phlegm–stasis obstruction syndrome	27	19.01%	97.18%
Spleen–kidney deficiency syndrome	2	1.41%	98.59%
Wind–cold obstructing collaterals syndrome	2	1.41%	100.00%
Total	142	100.0%

Figure 2.

Distribution of traditional Chinese medicine (TCM) syndrome types.

The distribution data revealed significant differences in the composition of TCM syndrome types among the 142 patients: dampness–heat obstruction syndrome was the most predominant type, with 82 cases, accounting for a high proportion of 57.75%: more than half of the total cases, making it the dominant syndrome type in this study population with RA and cerebral infarction. This was followed by liver–kidney deficiency syndrome and phlegm–stasis obstruction syndrome, which had identical distribution frequencies of 27 cases each, each accounting for 19.01%, and a combined proportion of 38.02%, representing the common syndrome types second only to dampness–heat obstruction. In contrast, cold–dampness obstruction syndrome, spleen–kidney deficiency syndrome, and wind–cold obstructing collaterals syndrome were extremely rare, with only 2 cases each, each accounting for 1.41%, and a cumulative proportion of only 4.23%, classifying them as the less common syndrome types in this study population.

3.1.1. Preliminary analysis of the distribution characteristics of less common syndromes

The very low prevalence of cold–dampness obstruction syndrome, spleen–kidney deficiency syndrome, and wind–cold obstructing collaterals syndrome in this study may be related to the following factors, considering the study design and disease characteristics.

1. Association between regional climate and disease pathology: the cases in this study were sourced from the East China region (characterized by a subtropical monsoon climate), which has a perennial humid and hot climate. Patients with RA complicated by cerebral infarction often present with long-term chronic inflammation (RA pathology) and vascular endothelial injury (cerebral infarction pathology). This dual pathological state is prone to generating “dampness–heat” pathogenic factors: dampness–heat accumulation in the joints exacerbates RA inflammation, while its obstruction of brain collaterals can induce or worsen cerebral infarction. Consequently, dampness–heat obstruction syndrome becomes the dominant type. In contrast, cold-natured syndrome types such as “cold–dampness” and “wind–cold” are less prevalent as they conflict with the regional climate and the febrile characteristics of the disease pathology.

2. Disease course and stage characteristics of the included population: this study included patients hospitalized “due to RA complicated by cerebral infarction,” most of whom were in the acute phase of the diseases (e.g., cerebral infarction onset ≤ 2 weeks, active RA stage). Spleen–kidney deficiency syndrome is more commonly seen in the chronic protracted stage of diseases (e.g., RA late stage with joint deformity accompanied by long-term malnutrition, post-cerebral infarction sequelae with consumption of qi and blood). Wind–cold obstructing collaterals syndrome typically occurs in the initial stage of acute Bi syndrome induced by exogenous wind–cold (its symptoms can easily be masked by neurological deficits after cerebral infarction, leading to clinical underdiagnosis). Therefore, the detection rate of these 2 syndrome types is low in an acutely ill, hospitalized population.

3. Stringency of syndrome differentiation criteria: the syndrome differentiation in this study required simultaneous fulfillment of the dual elements of “RA syndrome patterns” and “cerebral infarction syndrome patterns” (e.g., cold–dampness obstruction syndrome requires both “cold pain in the joints” and “cold congelation in the brain collaterals”). In clinical practice, however, most patients exhibit cold-natured or deficiency-natured syndromes in only 1 of the 2 diseases (either RA or cerebral infarction). Cases with overlapping dual syndrome patterns are inherently less common, thus resulting in a limited sample size for the ultimately included less common syndrome types.

It is important to note that all samples in this study were from East China. The humid-hot climatic background of this region is likely a significant environmental factor contributing to the absolute dominance of damp–heat obstruction syndrome. Therefore, the syndrome distribution pattern observed in this study, particularly the low proportion of rare syndrome types, may, to some extent, reflect regional characteristics. Caution is warranted when generalizing these conclusions to populations from regions with vastly different climates (e.g., cold-dry northern areas or humid southwestern areas).

3.2. Overall profile of influencing factors across different TCM syndrome types

The overall profile of influencing factors for patients with different TCM syndrome types is shown in Tables 3 and 4. Table 3 presents the overall measurement data, displaying relevant factors as means, while Table 4 presents the overall count data.

Table 3.

Overall status of patient measurement indicators.

	TCM syndrome types (number of examples)
	Cold–dampness obstruction syndrome (n = 2)	Dampness–heat obstruction syndrome (n = 82)	Phlegm–stasis obstruction syndrome (n = 27)	Liver–kidney deficiency syndrome (n = 27)	Spleen–kidney deficiency syndrome (n = 2)	Wind–cold obstructing collaterals syndrome (n = 2)
Gender (male/female)	0/2	15/67	6/21	4/23	0/2	1/1
Age (mean)	69.000	63.659	62.111	68.889	65.500	62.500
Hs-CRP (mean)	1.460	23.021	29.155	18.150	4.520	34.305
Glu (mean)	4.960	5.675	5.057	5.400	5.805	5.865
HDL-C (mean)	1.740	1.331	1.320	1.469	1.495	1.415
WBC (mean)	4.280	6.462	6.031	6.623	4.745	8.235
PLT (mean)	213.000	236.988	232.000	212.111	190.000	293.500
ANC (mean)	3.070	4.234	4.112	4.363	2.980	6.185
TG (mean)	1.095	1.561	1.111	1.559	1.100	1.480
Hb (mean)	131.500	113.915	110.333	118.556	115.500	122.000
RBC (mean)	4.100	3.876	3.760	3.915	3.630	3.805
TC (mean)	5.400	4.778	4.511	5.248	4.915	5.080
LDL-C (mean)	3.190	2.908	2.762	3.247	2.985	3.065

ANC = absolute neutrophil count, Hb = hemoglobin, Hs-CRP = high-sensitivity C-reactive protein, Glu = glucose, HDL-C = high-density lipoprotein cholesterol, LDL-C = low-density lipoprotein cholesterol, PLT = platelet count, RBC = red blood cell count, TC = total cholesterol, TG = triglyceride, WBC = white blood cell count.

Table 4.

Overall situation of patient count indicators.

	TCM syndrome types (number of examples)
	Cold–dampness obstruction syndrome (n = 2)	Dampness–heat obstruction syndrome (n = 82)	Phlegm–stasis obstruction syndrome (n = 27)	Liver–kidney deficiency syndrome (n = 27)	Spleen–kidney deficiency syndrome (n = 2)	Wind–cold obstructing collaterals syndrome (n = 2)
c-ANCA (normal/abnormal)	1/1	79/3	26/1	27/0	2/0	2/0
Anti-GBM (normal/abnormal)	2/0	80/2	26/1	26/1	2/0	2/0
Anti-MPO (normal/abnormal)	2/0	79/3	26/1	26/1	2/0	2/0
FR-ANCA (normal/abnormal)	2/0	79/3	25/2	25/2	2/0	2/0
FS-P-ANCA (normal/abnormal)	2/0	50/32	18/9	20/7	2/0	2/0
Anti-PR3 (normal/abnormal)	2/0	82/0	25/2	25/2	2/0	2/0
GPI (normal/abnormal)	2/0	42/40	18/9	22/5	2/0	2/0
AKA (normal/abnormal)	2/0	27/55	13/14	18/9	2/0	1/1
Anti-CCP IgG (normal/abnormal)	2/0	13/69	5/22	9/18	2/0	1/1
Anti-CENP-B (normal/abnormal)	2/0	79/3	27/0	26/1	2/0	2/0
CENP (normal/abnormal)	2/0	81/1	27/0	26/1	2/0	2/0
NDP (normal/abnormal)	2/0	82/0	27/0	26/1	2/0	1/1
Anti-dsDNA (normal/abnormal)	2/0	79/3	26/1	27/0	2/0	2/0
CFP (normal/abnormal)	2/0	77/5	26/1	24/3	2/0	2/0
CGP (normal/abnormal)	2/0	68/14	22/5	24/3	2/0	1/1
AHA (normal/abnormal)	2/0	78/4	26/1	26/1	2/0	2/0
HNP (normal/abnormal)	2/0	55/27	22/5	24/3	2/0	2/0
Anti-Jo1 (normal/abnormal)	2/0	78/4	24/3	26/1	2/0	2/0
NMP (normal/abnormal)	2/0	82/0	27/0	25/2	2/0	2/0
Anti-U1RNP/Sm (normal/abnormal)	2/0	78/4	23/4	26/1	2/0	1/1
NP (normal/abnormal)	2/0	75/7	25/2	21/6	2/0	2/0
ANA (normal/abnormal)	2/0	81/1	26/1	27/0	2/0	2/0
SNP (normal/abnormal)	2/0	56/26	17/10	18/9	2/0	1/1
Anti-Ro52 (normal/abnormal)	2/0	70/12	20/7	20/7	2/0	2/0
Anti-Ribo P (normal/abnormal)	2/0	77/5	26/1	270	2/0	2/0
Anti-Scl-70 (normal/abnormal)	2/0	78/4	27/0	27/0	2/0	2/0
Anti-Sm (normal/abnormal)	2/0	81/1	26/1	27/0	2/0	1/1
Anti-SSA (normal/abnormal)	2/0	69/13	21/6	22/5	2/0	2/0
Anti-SSB (normal/abnormal)	2/0	78/4	26/1	25/2	2/0	2/0

AHA = anti-histone antibody, AKA = anti-keratin antibody, ANA = anti-nucleosome antibody, anti-CCP IgG = anti-cyclic citrullinated peptide antibody IgG, anti-CENP-B = anti-centromere protein B antibody, anti-dsDNA = anti-double-stranded DNA antibody, anti-GBM = anti-glomerular basement membrane antibody, anti-Jo1 = anti-Jo-1 antibody, anti-MPO = anti-myeloperoxidase antibody, anti-PR3 = anti-proteinase 3 antibody, anti-Ribo P = anti-ribosomal P protein antibody, anti-Ro52 = anti-Ro-52 antibody, anti-Scl-70 = anti-topoisomerase I antibody, anti-Sm = anti-Smith antibody, anti-SSA = anti-Sjögren syndrome A antibody, anti-SSB = anti-Sjögren syndrome B antibody, anti-U1RNP/Sm = anti-U1 ribonucleoprotein/Smith antibody, c-ANCA = cytoplasmic anti-neutrophil cytoplasmic antibody, CENP = centromere pattern, CFP = cytoplasmic filamentous pattern, CGP = cytoplasmic granular pattern, FR-ANCA = formaldehyde-resistant ANCA, FS-P-ANCA = formaldehyde-sensitive P-ANCA, GPI = glucose-6-phosphate isomerase, HNP = homogeneous nuclear pattern, NDP = nuclear dots pattern, NMP = nuclear membrane pattern, NP = nucleolar pattern, SNP = speckled nuclear pattern.

3.2.1. Overall distribution characteristics of measurement data

3.2.1.1. Demographic and basic characteristics

Gender distribution: all syndrome types were predominantly female. The highest proportion of females was observed in the dampness–heat obstruction syndrome (67/82, 81.71%), with 15 males (18.29%). This was followed by liver–kidney deficiency syndrome (23 females, 85.19%; 4 males, 14.81%) and phlegm–stasis obstruction syndrome (21 females, 77.78%; 6 males, 22.22%). Among the less common syndromes, both cold–dampness obstruction syndrome and spleen–kidney deficiency syndrome cases were female (2/2 each), and wind–cold obstructing collaterals syndrome had 1 male and 1 female (1/1 each), consistent with the epidemiological feature of a higher overall prevalence of RA in females.

Age distribution: the mean age across syndrome types concentrated between 62 and 69 years. The highest mean age was found in liver–kidney deficiency syndrome (68.889 years), followed by cold–dampness obstruction syndrome (69.000 years, n = 2). The lowest mean age was in phlegm–stasis obstruction syndrome (62.111 years), while dampness–heat obstruction syndrome (63.659 years) and wind–cold obstructing collaterals syndrome (62.500 years, n = 2) were similar. This suggests that patients with liver–kidney deficiency syndrome may tend to be older, aligning with the TCM theory that the liver and kidney become deficient with age.

3.2.1.2. Inflammation-related indicators

hs-CRP: as a core inflammation marker in active RA, mean values differed significantly among syndromes. Wind–cold obstructing collaterals syndrome had the highest mean (34.305, n = 2), followed by phlegm–stasis obstruction syndrome (29.155) and dampness–heat obstruction syndrome (23.021). Cold–dampness obstruction syndrome had the lowest mean (1.460, n = 2), followed by spleen–kidney deficiency syndrome (4.520, n = 2) and liver–kidney deficiency syndrome (18.150). This suggests patients with phlegm–stasis obstruction syndrome and wind–cold obstructing collaterals syndrome may have more pronounced inflammatory activity, while cold–dampness obstruction syndrome (small sample size) had milder inflammation.

WBC and ANC: the trend for neutrophil count (an inflammatory cell indicator) was consistent with hs-CRP: highest in wind–cold obstructing collaterals syndrome (6.185, n = 2), with dampness–heat obstruction syndrome (4.234), liver–kidney deficiency syndrome (4.363), and phlegm–stasis obstruction syndrome (4.112) being similar; lowest in cold–dampness obstruction syndrome (3.070, n = 2). The overall distribution of WBC was similar to ANC, only wind–cold obstructing collaterals syndrome (8.235, n = 2) was significantly higher than other types, further supporting differences in inflammatory activity among syndromes.

3.2.1.3. Metabolism-related indicators

Glu: mean values for all syndromes were within the normal range (3.9–6.1 mmol/L), but distributions differed. Wind–cold obstructing collaterals syndrome had the highest mean (5.865, n = 2), followed by spleen–kidney deficiency syndrome (5.805, n = 2) and dampness–heat obstruction syndrome (5.675). Cold–dampness obstruction syndrome had the lowest mean (4.960, n = 2), with phlegm–stasis obstruction syndrome (5.057) and liver–kidney deficiency syndrome (5.400) in the middle. This suggests patients with dampness–heat, wind–cold, and spleen–kidney deficiency syndromes may require attention to Glu metabolism fluctuations.

Lipid indicators: HDL-C, “protective lipid”: cold–dampness obstruction syndrome had the highest mean (1.740 mmol/L, n = 2), followed by liver–kidney deficiency syndrome (1.469) and spleen–kidney deficiency syndrome (1.495, n = 2). Phlegm–stasis obstruction syndrome had the lowest mean (1.320), similar to dampness–heat obstruction syndrome (1.331), suggesting a potential “dyslipidemia tendency” with lower HDL-C in these patients.

LDL-C, “atherogenic lipid”: liver–kidney deficiency syndrome had the highest mean (3.247 mmol/L), followed by cold–dampness obstruction syndrome (3.190, n = 2) and wind–cold obstructing collaterals syndrome (3.065, n = 2). Phlegm–stasis obstruction syndrome had the lowest mean (2.762). This aligns with the TCM pathogenesis that “liver–kidney deficiency easily leads to endogenous phlegm–turbidity and blood stasis,” suggesting patients with this syndrome type may require enhanced vascular protection interventions.

3.2.1.4. Complete blood count indicators

Hb and RBC: mean Hb (reflecting anemia status) was highest in cold–dampness obstruction syndrome (131.500 g/L, n = 2), followed by wind–cold obstructing collaterals syndrome (122.000, n = 2) and liver–kidney deficiency syndrome (118.556). Phlegm–stasis obstruction syndrome had the lowest mean (110.333), similar to dampness–heat obstruction syndrome (113.915), suggesting a potential tendency for mild anemia (normal adult female 115–150 g/L, male 130–175 g/L) in these patients, possibly related to long-term chronic inflammatory consumption in RA.

PLT: wind–cold obstructing collaterals syndrome had the highest mean (293.500 × 10⁹/L, n = 2), followed by dampness–heat obstruction syndrome (236.988) and phlegm–stasis obstruction syndrome (232.000). Spleen–kidney deficiency syndrome had the lowest mean (190.000 × 10⁹/L, n = 2), and liver–kidney deficiency syndrome (212.111) was close to the normal range (125–350 × 10⁹/L), showing no clear trend of abnormal platelet aggregation or reduction.

The overall boxplots for the measurement data are shown in Figure 3.

Figure 3.

Overall distribution box plot of measurement indicators.

3.2.2. Overall distribution characteristics of count data (autoantibodies)

Count data are presented as binary classifications (“normal/abnormal”), focusing primarily on 3 categories: RA-specific antibodies, ANCA-associated antibodies, and other autoantibodies. Differences in abnormal rates among syndrome types were mainly concentrated in RA-related specific antibodies, while overall abnormal rates for ANCA and other autoantibodies were low.

3.2.2.1. RA-related specific antibodies (closely associated with diagnosis and disease activity)

Anti-CCP IgG: as a specific antibody for early RA diagnosis (specificity > 90%), abnormal rates were high across all syndromes. The highest abnormal rate was in dampness–heat obstruction syndrome (69/82, 84.15%), followed by phlegm–stasis obstruction syndrome (22/27, 81.48%) and liver–kidney deficiency syndrome (18/27, 66.67%). Among the less common syndromes, wind–cold obstructing collaterals syndrome had 1 abnormal case (1/2, 50%), while cold–dampness obstruction syndrome and spleen–kidney deficiency syndrome were both normal (2/2). This suggests patients with dampness–heat and phlegm–stasis obstruction syndromes are more likely to be positive for RA-specific antibodies, consistent with the “more obvious inflammatory activity” indicated by their measurement data.

AKA: the distribution of abnormal rates was similar to anti-CCP IgG: highest in dampness–heat obstruction syndrome (55/82, 67.07%), followed by phlegm–stasis obstruction syndrome (14/27, 51.85%) and liver–kidney deficiency syndrome (9/27, 33.33%). Among the less common syndromes, only wind–cold obstructing collaterals syndrome had 1 abnormal case (1/2), further supporting the association between dampness–heat obstruction syndrome and abnormal RA-specific antibodies.

GPI: as an RA activity marker, the highest abnormal rate was in dampness–heat obstruction syndrome (40/82, 48.78%), followed by phlegm–stasis obstruction syndrome (9/27, 33.33%) and liver–kidney deficiency syndrome (5/27, 18.52%). All less common syndromes were normal (2/2), corroborating the result suggested by hs-CRP that “inflammatory activity is more obvious in dampness–heat obstruction syndrome.”

3.2.2.2. ANCA-associated antibodies (indicators related to vasculitis and vascular injury)

Formalin-sensitive P-ANCA: the highest abnormal rate was in dampness–heat obstruction syndrome (32/82, 39.02%), followed by phlegm–stasis obstruction syndrome (9/27, 33.33%) and liver–kidney deficiency syndrome (7/27, 25.93%). The less common syndromes were all normal (2/2). Other ANCA antibodies (e.g., cytoplasmic ANCA [c-ANCA], anti-MPO) had very low overall abnormal rates (<5%); only dampness–heat obstruction syndrome had 3 cases (3/82, 3.66%) with abnormal c-ANCA. This suggests ANCA-associated antibody abnormalities might be relatively more common in dampness–heat obstruction syndrome but are not a main characteristic feature across all syndrome types.

3.2.2.3. Other autoantibodies (non-RA specific, reflecting overall autoimmune status)

Anti-Ro-52 antibody: the highest abnormal rate was in dampness–heat obstruction syndrome (12/82, 14.63%), followed by phlegm–stasis obstruction syndrome (7/27, 25.93%) and liver–kidney deficiency syndrome (7/27, 25.93%).

ANA and HNP: only dampness–heat obstruction syndrome had a few abnormal cases (1/82 and 2/82, respectively); other syndromes (including the less common ones) were predominantly normal.

Anti-Sm, anti-SSA, anti-Sjögren syndrome-related antigen B antibodies: Overall abnormal rates were < 5%, with no clear distribution pattern across syndromes, suggesting a weak association between these nonspecific autoantibodies and TCM syndrome types; they may not be indicators for differentiating syndromes in RA patients with cerebral infarction.

3.3. Statistical analysis of measurement data

3.3.1. Normality tests for measurement data

To ensure the appropriate selection of statistical methods for subsequent between-group comparisons (analysis of differences in measurement indicators among different TCM syndrome types), normality tests were 1st conducted on 12 core measurement indicators (including demographic, inflammatory, metabolic, and complete blood count indicators) from the 142 RA patients complicated with cerebral infarction. Considering the relatively large total sample size (n = 142) of this study and the need to verify the overall distribution characteristics of continuous data, the Kolmogorov–Smirnov test (K–S test) was used as the basis for judging normality, supplemented by Skewness and Kurtosis for auxiliary verification (Judgment criteria: P > .05 + |Skewness| < 1 + |Kurtosis| < 2 indicated conformity to a normal distribution; otherwise, it was considered non-normal). Detailed results are shown in Table 5.

Table 5.

Normal distribution test of measurement indicators.

Name	Mean	Standard deviation	Skewness	Kurtosis	Kolmogorov–Smirnov test
					Statistic D value	P
Age	64.444	9.465	0.044	-0.532	0.077	.041*
Hs-CRP	22.856	31.316	1.84	3.013	0.234	.000*
Glu	5.5	2.105	3.734	17.681	0.259	.000*
HDL-C	1.364	0.362	0.845	1.402	0.056	.329
LDL-C	2.952	0.786	0.615	0.338	0.078	.036*
TG	1.461	2.122	10.39	117.358	0.318	.000*
TC	4.831	1.173	0.906	1.372	0.082	.021*
WBC	6.381	2.293	1.106	1.862	0.082	.020*
RBC	3.86	0.461	-0.005	0.081	0.045	.698
Hb	114.5	15.716	-0.111	0.2	0.053	.428
ANC	4.229	1.971	1.355	3.396	0.107	.000*
PLT	231.106	84.759	0.991	1.255	0.109	.000*

ANC = absolute neutrophil count, Glu = glucose, Hb = hemoglobin, Hs-CRP = high-sensitivity C-reactive protein, HDL-C = high-density lipoprotein cholesterol, LDL-C = low-density lipoprotein cholesterol, PLT = platelet count, RBC = red blood cell count, TC = total cholesterol, TG = triglyceride, WBC = white blood cell count.

^*P < .05.

3.3.1.1. Normally distributed indicators (3 items)

HDL-C: K–S test P = .329 > .05, Skewness 0.845, Kurtosis 1.402; approximately normal.

RBC: K–S test P = .698 > .05, Skewness −0.005, Kurtosis 0.081; almost perfectly normal.

Hb: K–S test P = .428 > .05, Skewness −0.111, Kurtosis 0.200; slightly left-skewed but overall normal.

3.3.1.2. Non-normally distributed indicators (9 items)

Categorized by indicator type, all satisfied K–S test P < .05 and exhibited significant distribution deviation:

Inflammation-related (hs-CRP, absolute neutrophil count – ANC): hs-CRP P = .000 (Skewness 1.840), ANC P = .000 (Skewness 1.355); both moderately right-skewed.

Metabolism-related (Glu, TG, TC, LDL-C): TG showed the most significant deviation (P = .000, Skewness 10.390, Kurtosis 117.358). Glu P = .000 (Skewness 3.734); both severely right-skewed. TC (P = .021) and LDL-C (P = .036) were mildly right-skewed.

Complete blood count/demographics (WBC, PLT, age): WBC P = .020 (Skewness 1.106), PLT P = .000 (Skewness 0.991); both mildly right-skewed. Age P = .041 (Kurtosis −0.532); exhibited a mildly platykurtic (flat-topped) distribution.

3.3.2. Homogeneity of variance test

For the 3 indicators identified as normally distributed in Section 3.3.1 (HDL-C, RBC, and Hb), Levene test was employed to assess the homogeneity of variance across different TCM syndrome types (non-normal indicators will subsequently be analyzed using non-parametric tests, rendering variance homogeneity assessment unnecessary; their results are briefly presented here). The test criterion was: P > .05 indicates homogeneity of variance, and P < .05 indicates heterogeneity of variance. Specific results are shown in Table 6.

Table 6.

Results of the homogeneity of variance test.

	TCM syndrome type (standard deviation)						F	P
	Cold–dampness obstruction syndrome (n = 2)	Dampness–heat obstruction syndrome (n = 82)	Phlegm–stasis obstruction syndrome (n = 27)	Liver–kidney deficiency syndrome (n = 27)	Spleen–kidney deficiency syndrome (n = 2)	Wind–cold obstructing collaterals syndrome (n = 2)
Age	8.49	9.67	8.27	8.89	16.26	10.61	0.574	.72
Hs-CRP	1.32	28.47	38.56	33.1	4.29	43.88	1.202	.311
HDL-C	0.1	0.36	0.36	0.38	0.22	0.22	0.653	.66
LDL-C	0.21	0.78	0.77	0.81	1.15	0.43	0.634	.674
TC	0.23	1.18	1.02	1.29	1.21	0.71	0.71	.617
WBC	1.2	2.44	1.8	2.19	0.7	4.82	1.915	.096
RBC	0.25	0.44	0.62	0.35	0.34	0.42	2.868	.017*
Hb	7.78	15.65	19.05	12.11	6.36	1.41	2.493	.034*
ANC	1.29	2.13	1.42	1.82	1.85	4.59	1.548	.179
PLT	1.41	86.5	82.31	86.45	15.56	108.19	1.113	.356

ANC = absolute neutrophil count, Hb = hemoglobin, Hs-CRP = high-sensitivity C-reactive protein, LDL-C = low-density lipoprotein cholesterol, PLT = platelet count, RBC = red blood cell count, WBC = white blood cell count.

* indicates P < .05.

3.3.2.1. Homogeneity of variance results for normally distributed indicators

Among the 3 normally distributed indicators, 1 exhibited homogeneity of variance, while 2 exhibited heterogeneity:

HDL-C: Levene test F = 0.653, P = .66 > .05, indicating homogeneity of variance. The standard deviations across syndrome types fluctuated within a small range (0.10–0.38), suggesting consistent dispersion of HDL-C levels among different syndrome types. Consequently, standard 1-way analysis of variance (ANOVA) can be used for subsequent between-group comparisons.

RBC: Levene test F = 2.868, P = .017* < .05, indicating heterogeneity of variance. The standard deviations differed noticeably among syndrome types (0.25–0.62), with phlegm–stasis obstruction syndrome (0.62) showing the highest dispersion. Therefore, Brown–Forsythe ANOVA will be used subsequently to correct for the effects of heterogeneous variances.

Hb: Levene test F = 2.493, P = .034* < .05, indicating heterogeneity of variance. The standard deviations differed significantly among syndrome types (1.41–19.05), with high dispersion observed in phlegm–stasis obstruction syndrome (19.05) and dampness–heat obstruction syndrome (15.65). Similarly, Brown–Forsythe ANOVA will be employed for subsequent between-group comparisons.

3.3.2.2. Homogeneity of variance results for non-normally distributed indicators

For the remaining 9 non-normally distributed indicators, the P-values for Levene test were all > .05 (e.g., age P = .72, hs-CRP P = .311, PLT P = .356), confirming the non-normal distribution of these indicators. They will subsequently be analyzed using the non-parametric Kruskal–Wallis H test.

3.3.2.3. Core conclusion: determination of subsequent parametric test methods between syndrome types

HDL-C (normal + homogeneous variance) → standard 1-way ANOVA.

RBC, Hb (normal + heterogeneous variance) → Brown–Forsythe ANOVA.

Non-normal indicators → Kruskal–Wallis H test.

3.3.3. Between-group comparisons of measurement data across different TCM syndrome types

Based on the results of the homogeneity of variance tests in Section 3.3.2, between-group differences in measurement indicators across different TCM syndrome types were compared using standard 1-way ANOVA (for homogeneous variance), Brown–Forsythe ANOVA (for heterogeneous variance), or the Kruskal–Wallis H test (for non-normally distributed indicators). The results are presented in Table 7 (non-parametric test), Table 8 (Brown–Forsythe test), and Table 9 (1-way ANOVA), supplemented with effect size analysis (Table 10). The statistical significance threshold was set at P < .05.

Table 7.

Non-parametric rank sum test (Kruskal–Wallis) results.

	TCM syndrome type median M (P25, P75)						Kruskal–Wallis H	P
	Cold–dampness obstruction syndrome (n = 2)	Dampness–heat obstruction syndrome (n = 82)	Phlegm–stasis obstruction syndrome (n = 27)	Liver–kidney deficiency syndrome (n = 27)	Spleen–kidney deficiency syndrome (n = 2)	Wind–cold obstructing collaterals syndrome (n = 2)
Age	69.000 (63.0–75.0)	62.000 (55.8–71.0)	60.000 (55.0–69.0)	68.000 (64.0–78.0)	65.500 (54.0–77.0)	62.500 (55.0–70.0)	8.329	.139
hs-CRP	1.460 (0.5–2.4)	10.800 (1.4–34.0)	11.170 (3.3–36.6)	4.450 (0.6–20.1)	4.520 (1.5–7.5)	34.305 (3.3–65.3)	5.463	.362
Glu	4.960 (4.9–5.0)	5.005 (4.5–5.7)	4.760 (4.2–5.3)	4.930 (4.5–5.7)	5.805 (5.7–6.0)	5.865 (5.0–6.7)	6.757	.239
LDL-C	3.190 (3.0–3.3)	2.790 (2.3–3.4)	2.620 (2.2–3.2)	3.240 (2.7–3.9)	2.985 (2.2–3.8)	3.065 (2.8–3.4)	7.02	.219
TG	1.095 (1.0–1.2)	1.200 (0.8–1.6)	1.040 (0.7–1.3)	1.290 (1.1–1.9)	1.100 (0.7–1.5)	1.480 (1.0–2.0)	6.204	.287
TC	5.400 (5.2–5.6)	4.695 (4.0–5.4)	4.440 (3.7–5.0)	5.060 (4.5–6.1)	4.915 (4.1–5.8)	5.080 (4.6–5.6)	7.888	.163
WBC	4.280 (3.4–5.1)	6.300 (4.5–8.1)	6.080 (4.8–7.0)	5.890 (5.3–7.6)	4.745 (4.3–5.2)	8.235 (4.8–11.6)	4.666	.458
ANC	3.070 (2.2–4.0)	3.885 (2.7–5.3)	4.270 (3.0–4.8)	3.870 (3.3–5.3)	2.980 (1.7–4.3)	6.185 (2.9–9.4)	2.109	.834
PLT	213.000 (212.0–214.0)	215.000 (183.0–295.5)	233.000 (168.0–282.0)	193.000 (161.0–244.0)	190.000 (179.0–201.0)	293.500 (217.0–370.0)	3.933	.559

ANC = absolute neutrophil count, Glu = glucose, Hs-CRP = high-sensitivity C-reactive protein, LDL-C = low-density lipoprotein cholesterol, PLT = platelet count, TC = total cholesterol, TG = triglyceride, WBC = white blood cell count.

Table 8.

Brown–Forsythe ANOVA results.

	TCM syndrome type (mean ± standard deviation)						Brown F	P
	Cold–dampness obstruction syndrome (n = 2)	Dampness–heat obstruction syndrome (n = 82)	Phlegm–stasis obstruction syndrome (n = 27)	Liver–kidney deficiency syndrome (n = 27)	Spleen–kidney deficiency syndrome (n = 2)	Wind–cold obstructing collaterals syndrome (n = 2)
RBC	4.10 ± 0.25	3.88 ± 0.44	3.76 ± 0.62	3.91 ± 0.35	3.63 ± 0.34	3.80 ± 0.42	0.708	.627
Hb	131.50 ± 7.78	113.91 ± 15.65	110.33 ± 19.05	118.56 ± 12.11	115.50 ± 6.36	122.00 ± 1.41	2.645	.036*

Hb = hemoglobin, RBC = red blood cell count.

* indicates P < .05.

Table 9.

One-way ANOVA results.

	TCM syndrome type (mean ± standard deviation)						F	P
	Cold–dampness obstruction syndrome (n = 2)	Dampness–heat obstruction syndrome (n = 82)	Phlegm–stasis obstruction syndrome (n = 27)	Liver–kidney deficiency syndrome (n = 27)	Spleen–kidney deficiency syndrome (n = 2)	Wind–cold obstructing collaterals syndrome (n = 2)
HDL-C	1.74 ± 0.10	1.33 ± 0.36	1.32 ± 0.36	1.47 ± 0.38	1.50 ± 0.22	1.42 ± 0.22	1.171	.327

HDL-C = high-density lipoprotein cholesterol.

Table 10.

In depth analysis – effect measures (one-way ANOVA).

	SSB	SST	Partial η2	Cohen f
HDL-C	0.763	18.49	0.041	0.208

HDL-C = high-density lipoprotein cholesterol, SSB = Sjögren syndrome B.

3.3.3.1. 1-way ANOVA results (for HDL-C)

Only HDL-C met the “normal distribution + homogeneous variance” condition and was analyzed using standard 1-way ANOVA. The results showed:

The mean HDL-C across syndrome types ranged from 1.32 ± 0.36 (phlegm–stasis obstruction syndrome) to 1.74 ± 0.10 (cold–dampness obstruction syndrome), with the highest mean in cold–dampness obstruction syndrome and the lowest in phlegm–stasis obstruction syndrome.

No statistically significant difference was found between groups (F = 1.171, P = .327 > .05), suggesting that HDL-C levels were overall consistent across patients with different TCM syndrome types.

Further effect size analysis (Table 10) showed: partial η² = 0.041 (<0.06), Cohen f = 0.208 (<0.25), indicating that the syndrome type had weak explanatory power for HDL-C, and the between-group difference had limited practical clinical significance.

3.3.3.2. Brown–Forsythe ANOVA results (for RBC and Hb)

For RBC and Hb, which were “normally distributed but with heterogeneous variance,” the Brown–Forsythe test (a more robust method for correcting heterogeneous variances) was used. The results were as follows:

RBC: the mean RBC across syndrome types ranged from 3.63 ± 0.34 (spleen–kidney deficiency syndrome) to 4.10 ± 0.25 (cold–dampness obstruction syndrome). No statistically significant between-group difference was found (Brown F = 0.708, P = .627 > .05), suggesting no significant overall difference in RBC counts among patients with different syndrome types.

Hb: mean Hb levels differed noticeably among syndrome types, being highest in cold–dampness obstruction syndrome (131.50 ± 7.78 g/L) and lowest in phlegm–stasis obstruction syndrome (110.33 ± 19.05 g/L). A statistically significant between-group difference was found (Brown F = 2.645, P = .036* < .05), suggesting a potential association between TCM syndrome type and Hb level. Patients with cold–dampness obstruction syndrome tended to have relatively higher Hb levels, while those with phlegm–stasis obstruction syndrome had relatively lower levels.

3.3.3.3. Kruskal–Wallis H test results (for non-normal indicators)

The between-group comparison results for the 9 non-normally distributed indicators (Age, hs-CRP, Glu, LDL-C, TG, TC, WBC, ANC, PLT) showed:

The P-values for the Kruskal–Wallis test for all indicators were > .05 (range 0.139–0.834), indicating no statistically significant differences.

Although numerical differences existed in the means/medians of some indicators (e.g., the highest median age was 68.00 in liver–kidney deficiency syndrome, the lowest was 60.00 in phlegm–stasis obstruction syndrome; the highest median hs-CRP was 34.305 in wind–cold obstructing collaterals syndrome, the lowest was 4.450 in liver–kidney deficiency syndrome), the statistical results did not reach significance, likely due to the non-normal distribution of data and the small sample size of some syndrome types (n = 2). This suggests that the distribution of these indicators was generally consistent across different syndrome types.

3.3.3.4. Core conclusion

Among the measurement indicators compared across different TCM syndrome types, only Hb showed a statistically significant between-group difference (P = .036). The differences for all other indicators (including HDL-C, RBC, and the 9 non-normal indicators) were not statistically significant. This result suggests that Hb level may be associated with TCM syndrome types in RA patients complicated with cerebral infarction, while other inflammatory, metabolic, and routine blood indicators showed low discriminative power among the different syndrome types.

3.3.4. Post hoc multiple comparisons for Hb

Given that the Brown–Forsythe ANOVA for Hb in Section 3.3.3 revealed a statistically significant difference among the TCM syndrome types (Brown F = 2.645, P = .036), Tamhane T2 post hoc multiple comparison test was employed to further identify which specific pairs of syndrome types differed. This method is suitable for scenarios with heterogeneous variances and unequal sample sizes and offers greater robustness for small sample data. All pairwise comparisons among the 6 TCM syndrome types were conducted, with results detailed in Table 11. The statistical significance threshold remained at P < .05 for a statistically significant difference and P < .01 for a highly statistically significant difference.

Table 11.

Post hoc multiple comparison analysis results.

(I) TCM syndrome type	(J) TCM syndrome type	(I) Mean	(J) Mean	(I–J) Difference in mean value	P
CDOS	DHOS	131.5	113.915	17.585	.664
CDOS	PSOS	131.5	110.333	21.167	.399
CDOS	LKDS	131.5	118.556	12.944	.767
CDOS	SKDS	131.5	115.5	16	.644
CDOS	WCOCS	131.5	122	9.5	.907
DHOS	PSOS	113.915	110.333	3.581	.945
DHOS	LKDS	113.915	118.556	-4.641	.52
DHOS	SKDS	113.915	115.5	-1.585	1
DHOS	WCOCS	113.915	122	-8.085	.007*
PSOS	LKDS	110.333	118.556	-8.222	.332
PSOS	SKDS	110.333	115.5	-5.167	.971
PSOS	WCOCS	110.333	122	-11.667	.029*
LKDS	SKDS	118.556	115.5	3.056	.997
LKDS	WCOCS	118.556	122	-3.444	.717
SKDS	WCOCS	115.5	122	-6.5	.942

CDOS = cold–dampness obstruction syndrome, DHOS = dampness–heat obstruction syndrome, LKDS = liver–kidney deficiency syndrome, PSOS = phlegm–stasis obstruction syndrome, SKDS = spleen–kidney deficiency syndrome, WCOCS = wind–cold obstructing collaterals syndrome.

3.3.4.1. Core results of post hoc comparisons

Among the 12 pairwise comparisons, 2 pairs showed statistically significant differences, while the remaining 10 pairs showed no significant differences. The details are as follows:

Dampness–heat obstruction syndrome vs wind–cold obstructing collaterals syndrome

The mean Hb in wind–cold obstructing collaterals syndrome (122.00 g/L) was significantly higher than that in dampness–heat obstruction syndrome (113.915 g/L). The mean difference was −8.085 g/L (indicating wind–cold obstructing collaterals syndrome was 8.085 g/L higher), and the difference was highly statistically significant (P = .007 < .01). Considering the clinical context, patients with dampness–heat obstruction syndrome are often in the active stage of RA, where long-term chronic inflammation may lead to mild anemia (this syndrome had a lower mean Hb as mentioned in Section 3.2.1). In contrast, the wind–cold obstructing collaterals syndrome group had a small sample size (n = 2) and cases might be primarily induced by acute exogenous pathogens, implying relatively less inflammatory consumption and consequently higher Hb levels.

Phlegm–stasis obstruction syndrome vs wind–cold obstructing collaterals syndrome

The mean Hb in wind–cold obstructing collaterals syndrome (122.00 g/L) was significantly higher than that in phlegm–stasis obstruction syndrome (110.333 g/L). The mean difference was −11.667 g/L (indicating wind–cold obstructing collaterals syndrome was 11.667 g/L higher), and the difference was statistically significant (P = .029 < .05). The TCM pathogenesis of phlegm–stasis obstruction syndrome involves “intermingled phlegm and stasis obstructing qi and blood.” Impaired qi and blood circulation may affect erythrocyte production or metabolism, leading to relatively lower Hb levels. Wind–cold obstructing collaterals syndrome lacks significant phlegm–stasis obstruction, hence its relatively higher Hb levels.

3.3.4.2. Comparison pairs without statistical significance

The P-values for the other 10 pairwise comparisons were all > .05. The main characteristics were:

Comparisons involving cold–dampness obstruction syndrome: although cold–dampness obstruction syndrome had the highest mean Hb (131.50 g/L) and the largest numerical differences with phlegm–stasis obstruction syndrome (mean difference 21.167 g/L) and dampness–heat obstruction syndrome (mean difference 17.585 g/L), the very small sample size (n = 2) for this syndrome likely led to insufficient statistical power, resulting in P-values > .399 and no statistically significant differences.

Comparisons involving liver–kidney deficiency syndrome and spleen–kidney deficiency syndrome: the mean Hb levels in liver–kidney deficiency syndrome (118.556 g/L) and spleen–kidney deficiency syndrome (115.50 g/L) were moderate. The mean differences with other syndromes mostly ranged between 3 and 8 g/L, with P-values all > .332, indicating no statistical significance. This suggests these deficiency syndrome types did not exhibit specific differences in Hb levels.

Dampness–heat obstruction syndrome vs phlegm–stasis obstruction syndrome: although the mean Hb in phlegm–stasis obstruction syndrome (110.333 g/L) was lower than in dampness–heat obstruction syndrome (113.915 g/L), the mean difference was only 3.581 g/L, which is small numerically, resulting in P = .945 > .05 and no statistical significance.

3.3.4.3. Influencing factors and limitations

Impact of small sample sizes: the sample sizes for cold–dampness obstruction syndrome, spleen–kidney deficiency syndrome, and wind–cold obstructing collaterals syndrome were all n = 2. Excessively small sample sizes lead to inaccurate estimation of within-group variation, thereby reducing statistical power. Even if actual numerical differences exist between some groups (e.g., cold–dampness vs phlegm–stasis), achieving statistical significance is difficult. Larger sample sizes are needed for further validation.

Clinical variability of Hb: Hb levels are influenced by factors such as diet, nutritional status, and concomitant bleeding. This study did not fully control for these confounding factors, which might have masked or amplified some between-group differences. Future studies could include covariates like “nutritional scores” or “history of bleeding” for adjustment.

3.3.4.4. Core conclusion

The post hoc multiple comparisons for Hb revealed that only wind–cold obstructing collaterals syndrome had significantly higher Hb levels than both dampness–heat obstruction syndrome and phlegm–stasis obstruction syndrome. No statistically significant differences were found among other syndrome types. This result suggests that Hb level might have certain reference value for differentiating wind–cold obstructing collaterals syndrome from dampness–heat/phlegm–stasis obstruction syndromes.

3.4. Statistical analysis of count indicators

3.4.1. Univariate analysis of count indicators across different TCM syndrome types

To identify count indicators (primarily RA-related autoantibodies and ANCA-class antibodies) significantly associated with TCM syndrome types, Pearson chi-square test was used for univariate analysis. For cells with an expected frequency < 5, no Fisher exact test correction was applied since the total sample size (n = 142) was > 40. Statistical significance was defined as P < .05. The strength of association was evaluated using effect size measures (Cramer V, Phi coefficient, etc), with Cramer V < 0.15 indicating weak association, 0.15 to 0.3 moderate association, and >0.3 strong association. Detailed results are shown in Table 12 (chi-square test results) and Table 13 (effect size analysis results).

Table 12.

Chi-square test results.

	TCM syndrome types (number of examples)
	Cold–dampness obstruction syndrome (n = 2)	Cold–dampness obstruction syndrome (n = 2)	Cold–dampness obstruction syndrome (n = 2)	Cold–dampness obstruction syndrome (n = 2)	Cold–dampness obstruction syndrome (n = 2)	Cold–dampness obstruction syndrome (n = 2)	χ2	P
c-ANCA (normal/abnormal)	1/1	79/3	26/1	27/0	2/0	2/0	6.221	.285
Anti-GBM (normal/abnormal)	2/0	80/2	26/1	26/1	2/0	2/0	0.529	.991
Anti-MPO (normal/abnormal)	2/0	79/3	26/1	26/1	2/0	2/0	0.44	.994
FR-ANCA (normal/abnormal)	2/0	79/3	25/2	25/2	2/0	2/0	1.533	.909
FS-P-ANCA (normal/abnormal)	2/0	50/32	18/9	20/7	2/0	2/0	6.713	.243
Anti-PR3 (normal/abnormal)	2/0	82/0	25/2	25/2	2/0	2/0	7.925	.16
GPI (normal/abnormal)	2/0	42/40	18/9	22/5	2/0	2/0	14.759	.011
AKA (normal/abnormal)	2/0	27/55	13/14	18/9	2/0	1/1	16.59	.005
Anti-CCP IgG (normal/abnormal)	2/0	13/69	5/22	9/18	2/0	1/1	16.816	.005
Anti-CENP-B (normal/abnormal)	2/0	79/3	27/0	26/1	2/0	2/0	2.151	.828
CENP (normal/abnormal)	2/0	81/1	27/0	26/1	2/0	2/0	1.667	.893
NDP (normal/abnormal)	2/0	82/0	27/0	26/1	2/0	1/1	9.696	.084
Anti-dsDNA (normal/abnormal)	2/0	79/3	26/1	27/0	2/0	2/0	2.151	.828
CFP (normal/abnormal)	2/0	77/5	26/1	24/3	2/0	2/0	2.019	.846
CGP (normal/abnormal)	2/0	68/14	22/5	24/3	2/0	1/1	3.351	.646
AHA (normal/abnormal)	2/0	78/4	26/1	26/1	2/0	2/0	0.638	.986
HNP (normal/abnormal)	2/0	55/27	22/5	24/3	2/0	2/0	9.963	.076
Anti-Jo1 (normal/abnormal)	2/0	78/4	24/3	26/1	2/0	2/0	2.207	.82
NMP (normal/abnormal)	2/0	82/0	27/0	25/2	2/0	2/0	6.764	.239
ANTI-U1RNP/Sm (normal/abnormal)	2/0	78/4	23/4	26/1	2/0	1/1	6.4	.269
NP (normal/abnormal)	2/0	75/7	25/2	21/6	2/0	2/0	5.091	.405
ANA (normal/abnormal)	2/0	81/1	26/1	27/0	2/0	2/0	1.667	.893
SNP (normal/abnormal)	2/0	56/26	17/10	18/9	2/0	1/1	3.685	.596
Anti-Ro52 (normal/abnormal)	2/0	70/12	20/7	20/7	2/0	2/0	5.12	.401
Anti-Ribo P (normal/abnormal)	2/0	77/5	26/1	270	2/0	2/0	3.496	.624
Anti-Scl-70 (normal/abnormal)	2/0	78/4	27/0	27/0	2/0	2/0	4.477	.483
Anti-Sm (normal/abnormal)	2/0	81/1	26/1	27/0	2/0	1/1	6.952	.224
Anti-SSA (normal/abnormal)	2/0	69/13	21/6	22/5	2/0	2/0	2.841	.724
Anti-SSB (normal/abnormal)	2/0	78/4	26/1	25/2	2/0	2/0	1.01	.962

AHA = anti-histone antibody, AKA = anti-keratin antibody, ANA = anti-nucleosome antibody, anti-CCP IgG = anti-cyclic citrullinated peptide antibody IgG, anti-CENP-B = anti-centromere protein B antibody, anti-dsDNA = anti-double-stranded DNA antibody, anti-GBM = anti-glomerular basement membrane antibody, anti-MPO = anti-myeloperoxidase antibody, anti-PR3 = anti-proteinase 3 antibody, anti-Ribo P = anti-ribosomal P protein antibody, anti-Ro52 = anti-Ro-52 antibody, anti-Scl-70 = anti-topoisomerase I antibody, anti-SSA = anti-Sjögren syndrome A antibody, anti-SSB = anti-Sjögren syndrome B antibody, anti-U1RNP/Sm = anti-U1 ribonucleoprotein/Smith antibody, CENP = centromere pattern, CFP = cytoplasmic filamentous pattern, CGP = cytoplasmic granular pattern, FR-ANCA = formaldehyde-resistant ANCA, FS-P-ANCA = formaldehyde-sensitive P-ANCA, GPI = glucose-6-phosphate isomerase, HNP = homogeneous nuclear pattern, NDP = nuclear dots pattern, NMP = nuclear membrane pattern, NP = nucleolar pattern, SNP = speckled nuclear pattern.

Table 13.

In-depth analysis – effect size indicators.

Analysis items	Phi	Series connection coefficient	Corrected series connection coefficient	Cramer V
c-ANCA	0.312	0.298	0.422	0.312
Anti-GBM	0.051	0.051	0.072	0.051
Anti-MPO	0.04	0.04	0.057	0.04
FR-ANCA	0.096	0.095	0.135	0.096
FS-P-ANCA	0.184	0.181	0.256	0.184
Anti-PR3	0.217	0.212	0.3	0.217
GPI	0.295	0.283	0.4	0.295
AKA	0.325	0.309	0.437	0.325
Anti-CCP IgG	0.364	0.342	0.483	0.364
Anti-CENP-B	0.094	0.093	0.132	0.094
CENP	0.103	0.103	0.145	0.103
NDP	0.508	0.453	0.641	0.508
Anti-dsDNA	0.094	0.093	0.132	0.094
CFP	0.112	0.111	0.157	0.112
CGP	0.148	0.147	0.208	0.148
AHA	0.052	0.052	0.074	0.052
HNP	0.24	0.234	0.33	0.24
Anti-Jo1	0.123	0.122	0.173	0.123
NMP	0.247	0.239	0.339	0.247
Anti-U1RNP/Sm	0.258	0.25	0.354	0.258
NP	0.192	0.189	0.267	0.192
ANA	0.103	0.103	0.145	0.103
SNP	0.133	0.131	0.186	0.133
Anti-Ro52	0.172	0.169	0.239	0.172
Anti-Ribo P	0.124	0.123	0.174	0.124
Anti-Scl-70	0.146	0.144	0.204	0.146
Anti-Sm	0.407	0.377	0.533	0.407
Anti-SSA	0.115	0.114	0.162	0.115
Anti-SSB	0.073	0.073	0.103	0.073

AHA = anti-histone antibody, AKA = anti-keratin antibody, ANA = anti-nucleosome antibody, anti-CCP IgG = anti-cyclic citrullinated peptide antibody IgG, anti-CENP-B = anti-centromere protein B antibody, anti-dsDNA = anti-double-stranded DNA antibody, anti-GBM = anti-glomerular basement membrane antibody, anti-PR3 = anti-proteinase 3 antibody, anti-Ribo P = anti-ribosomal P protein antibody, anti-Ro52 = anti-Ro-52 antibody, anti-Scl-70 = anti-topoisomerase I antibody, anti-Sm = anti-Smith antibody, anti-SSA = anti-Sjögren syndrome A antibody, anti-SSB = anti-Sjögren syndrome B antibody, anti-U1RNP/Sm = anti-U1 ribonucleoprotein/Smith antibody, c-ANCA = cytoplasmic anti-neutrophil cytoplasmic antibody, CENP = centromere pattern, CFP = cytoplasmic filamentous pattern, CGP = cytoplasmic granular pattern, FR-ANCA = formaldehyde-resistant ANCA, FS-P-ANCA = formaldehyde-sensitive P-ANCA, GPI = glucose-6-phosphate isomerase, HNP = homogeneous nuclear pattern, NDP = nuclear dots pattern, NMP = nuclear membrane pattern, NP = nucleolar pattern, SNP = speckled nuclear pattern.

3.4.1.1. Count indicators significantly associated with TCM syndrome types (3 items)

Three count indicators showed P < .05 in the chi-square test and had effect sizes indicating “moderate or stronger association”:

GPI:

Chi-square test: χ² = 14.759, P = .011 < .05, indicating significant association with syndrome types.

Distribution across syndromes: the highest abnormality rate was in the damp–heat obstruction syndrome (40/82, 48.78%), followed by phlegm–stasis obstruction (9/27, 33.33%), and liver–kidney deficiency (5/27, 18.52%). Rare syndrome types (cold–damp, spleen–kidney deficiency, wind–cold obstruction) showed no abnormalities (0/2).

Effect size: Cramer V = 0.295 (moderate association), suggesting that GPI abnormality is moderately associated with TCM syndrome types. This aligns with quantitative indicators suggesting higher inflammatory activity in damp–heat obstruction syndrome (e.g., higher mean hs-CRP). As a specific marker for RA active phase, a high GPI abnormality rate indicates more active disease in this syndrome type.

AKA:

Chi-square test: χ² = 16.590, P = .005 < .01, indicating highly significant association with syndrome types.

Distribution across syndromes: the highest abnormality rate was in damp–heat obstruction (55/82, 67.07%), followed by phlegm–stasis obstruction (14/27, 51.85%), and liver–kidney deficiency (9/27, 33.33%). Among rare syndromes, only wind–cold obstruction had 1 abnormal case (1/2, 50%); the others were normal (0/2).

Effect size: Cramer V = 0.325 (moderate to strong association), the highest among the 3 indicators. AKA is an early diagnostic antibody for RA. Its high abnormality rate in damp–heat obstruction suggests that this syndrome may be more prone to RA-specific autoimmune reactions, consistent with the TCM pathogenesis of “damp–heat generating toxins and damaging joints and collaterals.”

Anti-CCP IgG:

Chi-square test: χ² = 16.816, P = .005 < .01, indicating highly significant association with syndrome types.

Distribution across syndromes: the highest abnormality rate was in damp–heat obstruction (69/82, 84.15%), followed by phlegm–stasis obstruction (22/27, 81.48%), and liver–kidney deficiency (18/27, 66.67%). Among rare syndromes, only wind–cold obstruction had 1 abnormal case (1/2, 50%); the others were normal (0/2).

Effect size: Cramer V = 0.364 (moderate to strong association), 2nd only to AKA. Anti-CCP IgG is the most specific antibody for RA (specificity > 90%). Its high abnormality rate in damp–heat and phlegm–stasis obstruction suggests a stronger association with RA autoimmune pathology, potentially serving as a core biomarker for these syndrome types in RA patients with cerebral infarction.

3.4.1.2. Count indicators not significantly associated with TCM syndrome types (27 items)

The remaining 27 count indicators had P > .05 in the chi-square test and fell into 2 categories:

ANCA-class antibodies: including cytoplasmic anti-neutrophil cytoplasmic antibody (c-ANCA, P = .285), anti-myeloperoxidase antibody (MPO-Ab, P = .994), and formaldehyde-sensitive P-ANCA (P = .243). All ANCA-class antibodies had low abnormality rates (<5%, e.g., c-ANCA abnormality in damp–heat obstruction was only 3/82, 3.66%) and effect sizes < 0.2 (e.g., MPO-Ab Cramer V = 0.04, weak association). This indicates that ANCA-related vasculitis indicators are weakly associated with TCM syndrome types in RA patients with cerebral infarction and are unlikely to be key differentiating indicators.

Other autoantibodies: including anti-glomerular basement membrane antibody (P = .991), anti-SSA antibody (P = .724), anti-Sm antibody (P = .224), and homogeneous nuclear antibody (P = .076). These are mostly non-RA-specific antibodies with generally low abnormality rates (<15%) and effect sizes mostly < 0.15 (e.g., anti-SSA antibody Cramer V = 0.115, weak association). This suggests no significant association with TCM syndrome types, making them impractical for syndrome differentiation in clinical practice.

3.4.1.3. Supplementary significance of effect size analysis

Effect size measures in Table 13 further validated the rationality of association strength:

The 3 significantly associated indicators all had Cramer V > 0.29 (moderate or stronger association) and corrected contingency coefficients > 0.4 (e.g., anti-CCP IgG corrected contingency coefficient = 0.483), indicating strong clinical relevance beyond mere statistical significance due to sample size.

Among non-significant indicators, only speckled nuclear antibody (Cramer V = 0.508) and anti-Sm antibody (Cramer V = 0.407) had high effect sizes. However, since their chi-square test P-values were > .05 (speckled nuclear P = .084, anti-Sm P = .224) and rare syndrome types had small sample sizes (n = 2) leading to data volatility, they were not considered truly associated with syndrome types.

3.4.1.4. Analysis limitations

Small sample size bias: rare syndrome types (cold–damp obstruction, spleen–kidney deficiency, wind–cold obstruction) each had n = 2. The “no abnormality” results for some antibodies may be due to insufficient sample size (e.g., GPI normality in rare syndromes may not reflect true distribution).

Theoretical frequency issue: some cells had expected frequencies < 5 (e.g., antibody abnormality counts in wind–cold obstruction were mostly 1 or 0). Although the large total sample size reduced error, this may still affect the stability of chi-square test results.

Uncontrolled confounding factors: factors such as medication history (e.g., immunosuppressants potentially reducing antibody abnormality rates) and disease course were not adjusted for, which may have obscured or amplified associations between indicators and syndrome types.

3.4.1.5. Core conclusions

Univariate chi-square tests revealed that GPI, AKA, and anti-CCP IgG were significantly associated with TCM syndrome types in RA patients with cerebral infarction (P < .05), all with moderate or stronger association strength. Damp–heat obstruction syndrome had the highest abnormality rates across these 3 indicators, suggesting their potential as biological markers for this syndrome type. The other 27 count indicators (e.g., ANCA-class antibodies, anti-SSA antibody) showed no significant association with syndrome types and are unlikely to be useful for syndrome differentiation in clinical practice.

3.4.2. Logistic regression analysis of different TCM syndrome types and related factors

(0)Rationale for choosing binary and multinomial logistic regression

The TCM syndrome types in this study represent a multicategorical outcome variable (6 syndrome types). However, a single regression method cannot comprehensively address the core question of “association between syndrome types and influencing factors.” Binary and multinomial logistic regression are complementary: the former focuses on the “overall association of a single syndrome type,” while the latter examines the “relative associations between syndrome types.” Their combination allows for a more systematic and precise exploration of the roles of influencing factors, for the following reasons:

Necessity of binary logistic regression

This method treats “1 specific syndrome type as the case group and all other syndrome types combined as the control group” (e.g., “damp–heat obstruction syndrome vs all other syndromes”). Its core advantage lies in focusing on independent influencing factors of a single syndrome type: particularly suitable for this study, where “multiple rare syndrome types exist” (cold–damp, spleen–kidney deficiency, and wind–cold obstruction syndromes each have n = 2). If directly included in a multinomial model, the small sample sizes of rare syndrome types may lead to unstable parameter estimates. In contrast, binary models enhance statistical power by “merging control groups,” making it easier to identify unique associations for rare or common syndrome types (e.g., the independent association of damp–heat obstruction syndrome with GPI and AKA).

Necessity of multinomial logistic regression

This method requires specifying a “reference syndrome type” (this study selected the clinically common and stable “phlegm–stasis obstruction syndrome” based on univariate results). Other syndrome types are compared separately with the reference group (e.g., “liver–kidney deficiency syndrome vs phlegm–stasis obstruction syndrome”). Its core advantage is quantifying the relative strength of associations between syndrome types: addressing questions such as “whether the effect of the same factor differs across syndrome types” (e.g., GPI may be a protective factor for liver–kidney deficiency syndrome but a risk factor for damp–heat obstruction syndrome). This avoids the issue of “masked differences between syndrome types” caused by “merging control groups” in binary models, providing more direct evidence for clinical syndrome differentiation.

1. Binary logistic regression results (Table 14).

Table 14.

Binary logistic regression analysis results.

	Regression coefficient	Standard error	z value	Wald χ2	P-value	OR value	95% CI
Liver–kidney deficiency syndrome vs other syndromes
GPI	-0.94	0.547	-1.718	2.952	.086	0.391	0.134–1.141
AKA	-0.849	0.502	-1.691	2.861	.091	0.428	0.160–1.144
Anti-CCP IgG	-0.074	0.527	-0.141	0.02	.888	0.929	0.331–2.607
HNP	-0.834	0.667	-1.251	1.564	.211	0.434	0.117–1.605
intercept	-0.574	0.416	-1.38	1.904	.168	0.563	0.249–1.273
Dampness–heat obstruction syndrome vs other syndromes
GPI	0.96	0.394	2.437	5.939	.015	2.611	1.207–5.651
AKA	0.796	0.396	2.01	4.042	.044	2.218	1.020–4.821
Anti-CCP IgG	0.375	0.469	0.801	0.642	.423	1.455	0.581–3.646
HNP	0.939	0.466	2.014	4.056	.044	2.558	1.025–6.382
intercept	-0.959	0.412	-2.33	5.427	.02	0.383	0.171–0.859
Phlegm–stasis obstruction syndrome vs other syndromes
GPI	-0.223	0.461	-0.483	0.233	.629	0.8	0.324–1.977
AKA	-0.243	0.466	-0.521	0.271	.602	0.784	0.315–1.956
Anti-CCP IgG	0.501	0.583	0.859	0.738	.39	1.651	0.526–5.177
HNP	-0.444	0.549	-0.809	0.655	.418	0.641	0.219–1.881
Intercept	-1.534	0.505	-3.037	9.225	.002	0.216	0.080–0.580

AKA = anti-keratin antibody, anti-CCP IgG = anti-cyclic citrullinated peptide antibody IgG, GPI = glucose-6-phosphate isomerase, HNP = homogeneous nuclear pattern.

Using “single syndrome type vs all other syndrome types” as the outcome variable, the independent associations of the 4 variables are as follows:

1. Damp–heat obstruction syndrome (case group) versus other syndromes (control group)

Three variables were independent risk factors for damp–heat obstruction syndrome (P < .05, OR > 1, 95% CI excluding 1):

GPI: regression coefficient = 0.96, P = .015, OR = 2.611 (95% CI: 1.207–5.651). This indicates that the probability of damp–heat obstruction syndrome in GPI-positive individuals is 2.611 times that of GPI-negative individuals, independent of other variables. Combined with TCM pathogenesis, GPI, as a marker of RA active phase, suggests active inflammation, aligning closely with the core pathogenesis of damp–heat obstruction syndrome: “damp–heat generating toxins and obstructing collaterals.”

AKA: regression coefficient = 0.796, P = .044, OR = 2.218 (95% CI: 1.020–4.821). The probability of damp–heat obstruction syndrome in AKA-positive individuals is 2.218 times that of negative individuals, further validating the independent association between RA-specific antibodies and damp–heat obstruction syndrome and suggesting that this syndrome type is more prone to autoimmune activation.

Homogeneous nuclear antibody: regression coefficient = 0.939, P = .044, OR = 2.558 (95% CI: 1.025–6.382). The probability of damp–heat obstruction syndrome in homogeneous nuclear antibody-positive individuals is 2.558 times that of negative individuals. Although this antibody is not RA-specific, its higher abnormality rate in damp–heat obstruction syndrome (27/82, 32.93%) may relate to the overall immune dysregulation in this syndrome type.

The intercept term P = .02 (OR = 0.383) indicates that after adjusting for the 4 variables, the baseline probability of damp–heat obstruction syndrome is low, and its diagnosis relies more on the positive status of the above 3 variables.

1. Liver–Kidney deficiency syndrome (case group) vs other syndromes (control group).

None of the variables reached statistical significance (P > .05), but trend associations were observed:

GPI (P = .086, OR = 0.391) and AKA (P = .091, OR = 0.428) both had OR values < 1 and approached significance, suggesting that GPI and AKA positivity may be “potential protective factors” for liver–kidney deficiency syndrome (i.e., GPI or AKA-positive individuals are more likely to exhibit other syndrome types) (e.g., damp–heat obstruction), while negative individuals are more likely to have liver–kidney deficiency syndrome. This aligns with the clinical feature of “relatively mild inflammatory activity in liver–kidney deficiency syndrome” (e.g., lower mean hs-CRP in quantitative indicators).

Anti-CCP IgG (P = .888) and homogeneous nuclear antibody (P = .211) showed no significant associations, indicating limited value in differentiating liver–kidney deficiency syndrome.

1. Phlegm–stasis obstruction syndrome (case group) vs other syndromes (control group).

None of the 4 variables showed statistical significance (P > .6), but the intercept term P = .002 (OR = 0.216) suggests:

The occurrence of phlegm–stasis obstruction syndrome is not independently associated with GPI, AKA, anti-CCP IgG, or homogeneous nuclear antibody. Its formation may relate to other unincorporated factors (e.g., disease duration, severity of vascular lesions).

The intercept OR = 0.216 (95% CI: 0.080–0.580) indicates that after adjusting for the 4 variables, the baseline probability of phlegm–stasis obstruction syndrome is significantly lower than that of other syndromes, possibly related to its chronic stage of “prolonged disease entering collaterals,” requiring further analysis with variables like disease duration.

1. Multinomial logistic regression results (Table 15)

Table 15.

Results of multi-classification logistic regression analysis.

Phlegm–stasis obstruction syndrome	Regression coefficient	Standard error	z value	Wald χ2	P-value	OR value	95% CI
HNP	-0.703	0.557	-1.262	1.592	.207	0.495	0.166–1.475
GPI	-0.574	0.472	-1.214	1.475	.225	0.563	0.223–1.422
AKA	-0.55	0.482	-1.142	1.305	.253	0.577	0.224–1.483
Anti-CCP IgG	0.119	0.621	0.192	0.037	.848	1.126	0.334–3.803
Intercept	-0.467	0.573	-0.815	0.664	.415	0.627	0.204–1.928

Liver–kidney deficiency syndrome	Regression coefficient	Standard error	z value	Wald χ2	P-value	OR value	95% CI
HNP	-1.099	0.681	-1.613	2.603	.107	0.333	0.088–1.266
GPI	-1.225	0.562	-2.178	4.746	.029	0.294	0.098–0.884
AKA	-1.051	0.519	-2.024	4.095	.043	0.35	0.126–0.968
Anti-CCP IgG	-0.423	0.575	-0.736	0.542	.462	0.655	0.213–2.020
Intercept	0.374	0.496	0.754	0.568	.451	1.453	0.550–3.841

AKA = anti-keratin antibody, anti-CCP IgG = anti-cyclic citrullinated peptide antibody IgG, GPI = glucose-6-phosphate isomerase, HNP = homogeneous nuclear pattern.

Using phlegm–stasis obstruction syndrome as the reference group (clinically common and stable in univariate results), independent influencing factors for other syndrome types relative to the reference group were analyzed. Only results for liver–kidney deficiency syndrome (a common syndrome type) are presented (rare syndrome types were excluded due to small sample sizes):

1. Reference group (phlegm–stasis obstruction syndrome)

None of the regression coefficients for the 4 variables were statistically significant (P > .2), further confirming that in the multinomial model, GPI, AKA, anti-CCP IgG, and homogeneous nuclear antibody are not independently associated with the occurrence of phlegm–stasis obstruction syndrome. The core influencing factors for this syndrome type require further exploration.

1. Liver–kidney deficiency syndrome (relative to phlegm–stasis obstruction syndrome)

Two variables were independent protective factors for liver–kidney deficiency syndrome relative to phlegm–stasis obstruction syndrome (P < .05, OR < 1):

GPI: regression coefficient = −1.225, P = .029, OR = 0.294 (95% CI: 0.098–0.884). This indicates that relative to phlegm–stasis obstruction syndrome, the probability of liver–kidney deficiency syndrome in GPI-positive individuals is 0.294 times that of negative individuals (i.e., GPI positivity more supports phlegm–stasis obstruction syndrome, while negativity more supports liver–kidney deficiency syndrome, consistent with the trend in binary regression).

AKA: regression coefficient = −1.051, P = .043, OR = 0.350 (95% CI: 0.126–0.968). Relative to phlegm–stasis obstruction syndrome, the probability of liver–kidney deficiency syndrome in AKA-positive individuals is 0.350 times that of negative individuals: AKA positivity can serve as a reference indicator for “excluding liver–kidney deficiency syndrome and favoring phlegm–stasis obstruction syndrome,” related to the difference in immune activity between the 2 syndromes (phlegm–stasis obstruction syndrome has slightly higher inflammatory activity).

Anti-CCP IgG (P = .462) and homogeneous nuclear antibody (P = .107) showed no significant associations, indicating that these variables cannot effectively differentiate between liver–kidney deficiency syndrome and phlegm–stasis obstruction syndrome.

• (3)Core conclusions and limitations
  • (a)Core conclusions

Damp–heat obstruction syndrome: GPI, AKA, and homogeneous nuclear antibody are its independent risk factors (all OR > 2), serving as potential diagnostic markers.

Liver–kidney deficiency syndrome: GPI and AKA are independent protective factors relative to phlegm–stasis obstruction syndrome (both OR < 0.4), useful for differentiation from phlegm–stasis obstruction syndrome.

Phlegm–stasis obstruction syndrome: no independent association with the 4 incorporated variables; its mechanism may relate to other factors such as disease duration or vascular lesions.

In a sensitivity analysis incorporating gender, the main conclusions regarding the significant associations of GPI and AKA with damp–heat obstruction syndrome remained robust (detailed data can be viewed in Data S1, Supplemental Digital Content, https://links.lww.com/MD/R346).

1. Limitations

Variable selection limitations: variables were screened based solely on univariate P < .1, without incorporating potential confounding factors such as clinical features (e.g., disease duration, medication history), possibly omitting important influencing factors.

Missing rare syndrome types: cold–damp obstruction, spleen–kidney deficiency, and wind–cold obstruction syndromes each had only 2 cases and were excluded from regression analysis, preventing evaluation of their influencing factors.

4. Discussion

This study focused on the distribution characteristics of TCM syndrome types and their associated influencing factors in patients with RA complicated by cerebral infarction. Through systematic statistical analysis (normality tests, analysis of variance, chi-square tests, and logistic regression), key findings were identified, including “damp–heat obstruction syndrome as the dominant syndrome type” and “GPI, AKA, and anti-CCP IgG as core indicators associated with syndrome types.” The following discussion interprets these results in depth by integrating TCM theory, modern medical mechanisms, and existing research, and analyzes the clinical significance, limitations, and future directions of this study.

4.1. Core interpretation of TCM syndrome type distribution characteristics

Among the 142 patients in this study, damp–heat obstruction syndrome had the highest proportion (57.75%), followed by liver–kidney deficiency syndrome and phlegm–stasis obstruction syndrome (19.01% each). The proportions of cold–damp obstruction syndrome, spleen–kidney deficiency syndrome, and wind–cold obstruction syndrome were extremely low (1.41% each). This distribution pattern is not coincidental but rather the result of the synergistic effects of regional climate, disease pathology, and TCM pathogenesis:

From a regional perspective, the cases in this study were sourced from East China (a subtropical monsoon climate), characterized by year-round high temperature and humidity, making the body susceptible to external dampness. Patients with RA complicated by cerebral infarction have a dual pathological state of “RA chronic inflammation” and “cerebral infarction-induced vascular endothelial damage” – inflammatory factors such as tumor necrosis factor-α (TNF-α) and interleukin-6 (IL-6) released during active RA inflammation can promote the generation of “internal dampness.” The combination of internal and external dampness, stagnating and transforming into heat, forms the “damp–heat” pathogenesis. Consequently, damp–heat obstruction syndrome becomes the dominant syndrome type. This aligns with previous meta-analysis results indicating a high incidence of damp–heat syndrome in RA patients from East China (Huang Jing et al meta-analysis of RA patients showed that the proportion of damp–heat obstruction syndrome in RA patients in East China was significantly higher than in northern regions).^[33]

From the perspective of disease stage, all patients included in this study were in the acute phase, hospitalized for “RA complicated by cerebral infarction” (cerebral infarction onset ≤ 2 weeks, RA active phase): Damp–heat obstruction syndrome often corresponds to the acute phase of the disease, where damp–heat generates toxins, obstructing the joints and brain collaterals, thereby exacerbating RA joint inflammation and inducing the progression of cerebral infarction. Liver–kidney deficiency syndrome and phlegm–stasis obstruction syndrome often represent an “acute phase superimposed on a chronic underlying pathogenesis” – prolonged RA mean disease duration 8.2 years in this study easily consumes and damages liver and kidney essence and blood (liver–kidney deficiency), while inflammation-mediated vascular endothelial damage promotes the generation of phlegm and stasis (phlegm–stasis obstruction). Rare syndrome types (cold–damp, spleen–kidney deficiency, wind–cold) have very low detection rates because they contradict the “heat manifestations of the acute phase” and the “regional damp–heat” (cold–damp and wind–cold belong to cold-natured syndromes) or correspond to chronic disease stages (spleen–kidney deficiency is more common in the late stages of RA). This is consistent with the conclusions of Wu Jiaqi et al 2024 study on the distribution patterns of RA syndrome types.^[34]

4.2. Clinical and mechanistic significance of quantitative indicator associations

Among the quantitative indicators in this study, only Hb showed intergroup differences: Hb levels in the wind–cold obstruction syndrome were significantly higher than those in the damp–heat obstruction syndrome (P = .007) and the phlegm–stasis obstruction syndrome (P = .029). No statistical differences were found for other indicators (e.g., hs-CRP, blood lipids, routine blood tests). This result needs to be interpreted through the lens of both TCM pathogenesis and modern medical mechanisms.

From a TCM perspective, Hb levels are closely related to the “generation of qi and blood”: The core pathogenesis of phlegm–stasis obstruction syndrome is “intermingled phlegm and stasis, obstructing qi and blood.” Internal blood stasis prevents the generation of new blood, and exuberant phlegm turbidity impairs spleen transport function (the spleen is the source of qi and blood generation), hence the lowest Hb levels. In damp–heat obstruction syndrome, “damp–heat consumes qi and blood” – damp–heat toxins can scorch the collaterals, leading to occult blood loss, while damp–heat encumbering the spleen affects qi and blood generation, resulting in lower Hb levels. Wind–cold obstruction syndrome is mostly “acute onset induced by external contraction of wind–cold,” characterized by short disease duration and mild consumption, with qi and blood not significantly impaired, hence higher Hb levels.

From a modern medical perspective, RA patients commonly exhibit “anemia of chronic disease,” whose mechanism is related to inflammatory cytokines (such as IL-6) inhibiting erythropoiesis and shortening red blood cell lifespan (IL-6 can inhibit iron absorption and release by promoting hepatic hepcidin synthesis and directly inhibit the proliferation of bone marrow erythroid precursor cells): In this study, the mean hs-CRP levels in damp–heat obstruction syndrome and phlegm–stasis obstruction syndrome (23.021, 29.155 mg/L) were notably higher than in wind–cold obstruction syndrome (although not statistically different, the value reached 34.305 mg/L, possibly masked by the small sample size), suggesting more active inflammatory activity in the first 2 syndrome types, which may lead to decreased Hb through the aforementioned mechanisms. The wind–cold obstruction syndrome had a small sample size (n = 2) and might lack long-term chronic inflammatory accumulation, hence Hb was not significantly affected.

It is important to note that although other quantitative indicators (e.g., hs-CRP, LDL-C) showed no statistical differences, numerical trends existed (e.g., phlegm–stasis obstruction syndrome had the highest hs-CRP). This might be related to “low statistical power due to small sample sizes of rare syndrome types” and “failure to control for confounding factors like disease duration and medication history.” More associations might be discovered upon expanding the sample size in future studies.

4.3. Value of count indicators and logistic regression results for syndrome differentiation

The most critical finding of this study is the significant association of GPI, AKA, and anti-CCP IgG with TCM syndrome types. They were identified as independent risk factors for damp–heat obstruction syndrome (binary logistic regression: OR = 2.611, 2.218, both P < .05) and simultaneously as independent protective factors for liver–kidney deficiency syndrome relative to phlegm–stasis obstruction syndrome (multinomial logistic regression: OR = 0.294, 0.350, both P < .05). This provides a key basis for the “objective differentiation of syndromes” in RA complicated by cerebral infarction.

4.3.1. Mechanistic logic of the indicator–syndrome association

GPI, AKA, and anti-CCP IgG are all core indicators for RA diagnosis and disease activity assessment. Their association with syndrome types essentially reflects the “correspondence between TCM syndrome types and RA immune-inflammatory activity”:

GPI: as a specific marker for the active phase of RA, its level positively correlates with RA joint swelling and tenderness scores (Chen Yalin et al 2022 study showed that the GPI-positive group had significantly higher swollen joint count, tender joint count, erythrocyte sedimentation rate, CRP, and Disease Activity Score 28 (DAS28) than the negative group, and GPI levels directly correlated with the degree of synovitis). In this study, damp–heat obstruction syndrome had the highest GPI abnormality rate (48.78%), and it was an independent risk factor, suggesting that damp–heat obstruction syndrome is associated with the most active RA immune-inflammatory activity – the TCM pathogenesis of “damp–heat generating toxins” might correspond to RA synovial cell proliferation and immune cell infiltration (e.g., T cell, macrophage activation). Toxins damaging collaterals further induce cerebral infarction, forming a vicious cycle of “inflammation–syndrome type–disease progression.”

AKA and anti-CCP IgG: Both are highly specific antibodies for the early diagnosis of RA (anti-CCP IgG specificity > 90%). Korkmaz C et al 2006 study showed that anti-CCP antibodies have a specificity of 95.6% for RA diagnosis, and their positivity is closely related to the degree of RA bone erosion. MATSUMOTO et al 2006 study further confirmed that the positive rate of anti-CCP antibodies is significantly higher in RA patients during the active phase, serving as a predictive indicator for RA progression. In this study, the abnormality rates of AKA and anti-CCP IgG in damp–heat obstruction syndrome (67.07%, 84.15%) were significantly higher than in liver–kidney deficiency syndrome (33.33%, 66.67%), and they were independent risk factors for damp–heat obstruction syndrome. This indicates that damp–heat obstruction syndrome is more prone to be accompanied by the activation of RA-specific autoimmune responses – TCM “damp–heat toxins” might promote the synergistic progression of RA and cerebral infarction by damaging articular cartilage (corresponding to exposure of keratin antigens recognized by AKA) and promoting synovitis (corresponding to the generation of citrullinated proteins recognized by anti-CCP).

4.3.2. Clinical utility for syndrome differentiation

The multinomial logistic regression results further clarify that GPI and AKA can be used to differentiate between liver–kidney deficiency syndrome and phlegm–stasis obstruction syndrome (relative to phlegm–stasis obstruction syndrome, the probability of having liver–kidney deficiency syndrome is reduced by more than 60% in GPI or AKA-positive individuals). This conclusion holds significant clinical value.

The clinical symptoms of “phlegm–stasis obstruction syndrome” and “liver–kidney deficiency syndrome” in RA patients with cerebral infarction can easily be confused (e.g., both may present with joint pain and dizziness), but the treatment strategies differ significantly – phlegm–stasis obstruction syndrome requires “activating blood, resolving stasis, transforming phlegm, and unblocking collaterals” (e.g., using Dan Shen Yin combined with Wen Dan Tang), while liver–kidney deficiency syndrome requires “tonifying the liver and kidney, replenishing essence, and filling the marrow” (e.g., using Du Huo Ji Sheng Tang). GPI and AKA testing is convenient (available in routine laboratories) and can serve as “auxiliary indicators for syndrome differentiation”: if a patient is positive for GPI and/or AKA, phlegm–stasis obstruction syndrome is more likely; if negative, liver–kidney deficiency syndrome should be considered more cautiously, avoiding treatment deviations due to incorrect syndrome differentiation.

Furthermore, the lack of association between phlegm–stasis obstruction syndrome and all included indicators suggests that the core influencing factors for this syndrome type might be variables not included in this analysis, such as “disease duration” and “severity of vascular lesions” (e.g., carotid plaque thickness) – phlegm and stasis are the results of “prolonged disease entering the collaterals,” and their formation is related to long-term inflammation-induced vascular remodeling and platelet activation. Master of TCM Lei Zhongyi proposed that “phlegm–stasis intermingling is the core pathogenesis of atherosclerosis.” Fan Chaomei 2023 research also confirmed that the pathological processes of atherosclerosis, such as lipid deposition and fibrous proliferation, highly align with the TCM description of “phlegm turbidity and static blood obstructing the collaterals,” with chronic inflammation being the common pathological basis for both. Subsequent studies need to incorporate such variables for further exploration.

The strong association of “damp–heat obstruction syndrome” with RA-specific immune activation indicators (GPI, AKA, anti-CCP IgG) in this study suggests that the TCM pathogenesis of “damp–heat congelation obstructing meridians” may be intrinsically linked to active autoimmune responses, synovitis, and related cytokine storms (e.g., high expression of TNF-α, IL-6, IL-17) in RA. Similarly, the association of “phlegm–stasis obstruction syndrome” with relatively lower Hb levels may reflect the processes of “consumption of qi and blood” and “endogenous blood stasis” in a chronic inflammatory state, involving pathological aspects such as vascular endothelial injury, platelet activation, and fibrinolysis system abnormalities. Future mechanistic studies need to directly measure serum cytokine profiles, immune cell subsets (e.g., Th17/Treg balance), and vascular endothelial function markers in these patients to interpret the biological basis of TCM syndromes at the molecular level.

4.4. Consistency with and innovation beyond existing research

The findings of this study show both consistency with and innovation beyond existing research:

Consistency: the results are consistent with studies on syndrome types in RA patients alone, which also identified “damp–heat obstruction syndrome as a common syndrome type” and a “positive correlation between anti-CCP IgG and damp–heat syndrome.” The results of Wu Jiaqi et al study showed a positive rate of 86.0% for anti-CCP antibodies in RA patients during the active phase, with the highest expression level found in the damp–heat syndrome type. This is close to the 84.15% abnormality rate of anti-CCP IgG in the damp–heat obstruction syndrome in our study. This indicates that the core syndrome characteristics of RA (such as the association between damp–heat and immune activity) remain stable even after complicating with cerebral infarction, providing evidence for the “continuity of syndrome differentiation in RA complicated by other diseases.”

Innovation: this study is the 1st to focus on the specific population of “RA complicated by cerebral infarction,” clarifying the association between syndrome types and cerebral infarction-related indicators (such as Hb). Although Fujibayashi et al study focused on cerebrovascular events in RA patients, it did not involve TCM syndrome type analysis. This study, by employing the complementary methods of “binary + multinomial logistic regression,” overcame the limitation of “small sample sizes for rare syndrome types” and more precisely explored the value of indicators for syndrome differentiation. Previous studies mostly analyzed the syndrome types of RA or cerebral infarction separately, without focusing on the syndrome characteristics after their co-occurrence. This study fills that gap.

4.5. Study limitations

This study has the following limitations, which need to be addressed in future research:

Sample size and geographical limitations: the sample sizes for cold–dampness obstruction syndrome, spleen–kidney deficiency syndrome, and wind–cold obstructing collaterals syndrome were extremely small (only 2 cases each). While this objectively reflects the natural distribution under our study’s regional climate and strict diagnostic criteria, it severely limits the statistical power for comparisons and the reliability of independent conclusions regarding these specific syndromes. Although we employed strategies such as merging control groups in regression analyses to explore the major syndromes, the characteristics of these rare syndrome types and their associations with objective indicators require future validation through extended recruitment periods and, most importantly, multi-center collaborative studies to accumulate sufficient samples (e.g., achieving ≥ 20 cases per syndrome type) for in-depth analysis.

Furthermore, all cases were sourced from East China, where the damp–heat climate influences syndrome distribution, making it difficult to generalize the results to cold and dry northern regions. This was a single-center study, and all cases were from East China. The influence of regional humid-hot climate on syndrome distribution is pronounced, meaning the conclusions may not be generalizable to other regions with dry or cold climates. Future multi-center studies including cases from different climatic zones (e.g., northern, southwestern China) are needed to verify and refine the generalizability of the TCM syndrome distribution pattern in RA patients complicated with cerebral infarction. Huang Jing et al meta-analysis has confirmed significant regional differences in RA syndrome distribution; for example, the proportion of cold–damp obstruction syndrome in Southwest China is as high as 58%, far exceeding the 1.41% in this study. Subsequent research requires multi-center, large-sample studies that include cases from different regions.

As a retrospective study, our control over potential confounding factors (such as precise RA duration, detailed medication dosage/duration, and exact NIHSS scores) was limited. Although our analysis focused on core indicators, residual confounding effects from these incompletely adjusted clinical factors cannot be ruled out. Future prospective studies should systematically collect this information and employ more rigorous methods such as propensity score matching or multivariate adjustment models to control for confounders and verify the independence of the preliminary associations found in this study.

This study primarily explored associations at the phenomenological level based on routine clinical lab indicators. Although we inferred potential immune-inflammatory mechanisms in the discussion based on the literature, we did not directly measure key inflammatory cytokines (e.g., TNF-α, IL-6, IL-17) or perform immune cell profiling. Therefore, this study cannot provide experimental evidence for the direct correspondence between syndromes like “damp–heat” or “phlegm–stasis” and specific biological pathways, which is a core question for future translational research.

Insufficient control for confounding factors: confounding factors such as “disease duration,” “RA medication history” (e.g., methotrexate and biologics may affect inflammatory markers and antibody levels), and “severity of cerebral infarction” (e.g., NIHSS score) were not included. This may have obscured or amplified the associations between indicators and syndrome types. Future studies need to further control for these factors using propensity score matching or multivariate adjustment.

Lack of mechanistic and prognostic exploration: this is a cross-sectional study. It did not deeply explore the association between syndrome types and immune mechanisms (e.g., Th17, Treg cells), nor did it analyze the relationship between syndrome types and patient prognosis (e.g., cerebral infarction recurrence rate, RA joint function deterioration). Therefore, it cannot provide evidence for “prognosis guided by syndrome differentiation.” Subsequent prospective cohort studies are needed.

The cross-sectional nature of this study means it can only reveal associations, not confirm causality, nor evaluate the prognostic predictive value of syndromes or indicators. This is one of the major limitations inherent to the study design.

Furthermore, this study is an exploratory association analysis. Due to sample size limitations (particularly the highly disproportional sample sizes of target syndromes, e.g., only 2 cases of wind–cold obstructing collaterals syndrome), diagnostic test characteristics analyses such as receiver operating characteristic curve analysis to determine diagnostic thresholds were not performed at this stage. Therefore, the indicator differences reported (e.g., Hb levels) currently suggest a statistical association and potential trend but cannot yet be used as quantitative cutoff values for clinical diagnosis.

4.6. Clinical significance and future prospects

4.6.1. Clinical significance

The clinical value of this study’s findings is mainly reflected in 2 aspects:

Objectification of syndrome differentiation: it identifies GPI, AKA, and anti-CCP IgG as potential “auxiliary diagnostic indicators” for damp–heat obstruction syndrome, and Hb can assist in differentiating wind–cold obstruction syndrome from damp–heat/phlegm–stasis obstruction syndromes. This reduces the subjectivity of “differentiating syndromes based solely on symptoms,” which is particularly beneficial for less experienced clinicians.

Treatment precision: based on the indicators associated with syndrome types, treatment plans can be adjusted more targetedly. For example, if patients with damp–heat obstruction syndrome are positive for GPI and anti-CCP IgG, it suggests active immune inflammation. Anti-inflammatory treatment (e.g., NSAIDs) could be combined with the foundational approach of clearing heat and eliminating dampness. For patients with phlegm–stasis obstruction syndrome and decreased Hb, supplementing qi and nourishing blood (e.g., adding Astragalus membranaceus, Angelica sinensis) should be considered alongside activating blood and resolving stasis, thereby improving treatment effectiveness.

TCM emphasizes “treatment based on syndrome differentiation.” Misdiagnosis of the syndrome leads directly to fundamental errors in therapeutic principles and herbal prescription selection, potentially rendering treatment ineffective or even aggravating the condition. For example, if a patient with “cold–dampness obstruction syndrome” is misdiagnosed as having “damp–heat obstruction syndrome,” the inappropriate use of large doses of heat-clearing and dampness-drying herbs could damage spleen and stomach yang, causing the cold–damp pathogen to become entrenched and joint pain to persist and worsen. Conversely, misdiagnosing “damp–heat obstruction syndrome” as a deficiency or cold syndrome and overusing warming and tonifying herbs would be akin to “adding fuel to the fire,” potentially intensifying heat and inflammatory activity, and even affecting cerebrovascular stability. Therefore, seeking auxiliary objective indicators aims to reduce the subjective variability in syndrome differentiation, minimizing the risk of misjudging “cold, heat, deficiency, excess” from the source, thereby ensuring treatment safety and efficacy. The specific associations between indicators and syndrome types suggested by this study provide preliminary, quantifiable reference clues for this purpose.

4.6.2. Future prospects

Future research can advance in 3 directions:

Sample expansion and multi-center validation: include large-sample cases from multiple regions across the country to validate the universality of the syndrome distribution pattern. Simultaneously, increase the sample size of rare syndrome types to enhance the reliability of the results.

This study preliminarily identified associations between specific indicators (e.g., Hb, GPI, AKA) and syndrome types. An important direction for future research is to evaluate the diagnostic performance of these promising indicators in larger, independent, multi-center cohorts. For instance, receiver operating characteristic curve analysis could be employed to explore optimal cutoff values with the best sensitivity and specificity for indicators such as Hb level in differentiating “wind–cold obstructing collaterals syndrome” from “damp–heat/phlegm–stasis obstruction syndromes,” or for GPI and AKA levels in assisting the diagnosis of “damp–heat obstruction syndrome.” This would lay the groundwork for developing objective, quantitative auxiliary diagnostic tools for TCM syndrome differentiation.

Third, and most significant for clinical practice, is the conduct of prospective, multi-center cohort studies. Such studies should systematically assess patients’ TCM syndrome types and the series of objective indicators suggested by this research at baseline, followed by long-term follow-up (e.g., 1–3 years). Primary endpoints should include cerebral infarction recurrence, new vascular events, RA disease activity progression (e.g., change in DAS28 score), radiographic joint progression, and functional status changes. Only through such a design can it be truly verified whether specific syndrome types (e.g., damp–heat obstruction syndrome) or their corresponding objective indicator profiles can independently predict adverse outcomes. This would provide high-level evidence for implementing “treatment based on syndrome differentiation” interventions to improve long-term prognosis and may reveal potential causal relationships.

In-depth mechanistic exploration: utilize technologies like transcriptomics and metabolomics to analyze differences in immune cells (e.g., Th17, Treg cells) and inflammatory cytokines (e.g., IL-17, TNF-α) among patients with different syndrome types, revealing the molecular mechanisms behind “heightened immune activity in damp–heat obstruction syndrome” (research by Chen Yalin et al has confirmed that GPI levels are related to the degree of synovitis and cytokine release, which can serve as an entry point for mechanistic studies).

Prognostic and intervention research: conduct prospective studies to observe the cerebral infarction recurrence rate and changes in RA joint function scores in patients with different syndrome types, validating the effectiveness of “treatment based on syndrome differentiation in improving prognosis.” This will provide high-level evidence for the integrated Chinese and Western medicine treatment of RA complicated by cerebral infarction.

Upon accumulating a sufficiently large-sample size, future research can move beyond single-indicator association analysis towards constructing multi-indicator combined diagnostic models. For example, indicators such as GPI, AKA, anti-CCP IgG, Hb, and potentially novel biomarkers (e.g., specific cytokines) could be integrated. Multivariate logistic regression, LASSO regression, or machine learning algorithms (e.g., random forest, support vector machines) could be employed to build syndrome classification prediction models. By using cross-validation and independent external validation, the discrimination accuracy, sensitivity, and specificity of such combined models could be evaluated and optimized. This effort holds the promise of eventually developing digital tools to aid TCM syndrome differentiation for clinical decision support.

Finally, with the advancement of artificial intelligence technology, its application in TCM research holds great promise. For instance, future studies could explore using natural language processing techniques to automatically and standardizedly extract TCM diagnostic information (e.g., symptoms, tongue and pulse descriptions) from both structured and unstructured electronic medical record text. Such an approach has the potential to reduce the subjective bias and labor intensity associated with manual data extraction in retrospective studies, improving the efficiency and consistency of large-scale data processing. However, its successful application depends on high-quality, standardized clinical text data for training, which requires further synergy between TCM terminology standardization and clinical documentation norms. The data generated in this study through rigorous manual adjudication provides a reliable benchmark dataset for future exploration of such AI-assisted methods.

In summary, this study clarifies the syndrome distribution characteristics of “damp–heat obstruction syndrome being the main type” in patients with RA complicated by cerebral infarction. It also identifies GPI, AKA, anti-CCP IgG, and Hb as key indicators for syndrome differentiation and auxiliary diagnosis, providing an evidence-based foundation for the precise TCM diagnosis and treatment of these patients. Further larger-scale and more in-depth research is needed to refine the syndrome type system and promote the application of integrated Chinese and Western medicine in the diagnosis and treatment of RA complicated by cerebral infarction.

Author contributions

Conceptualization: Ji Yang.

Data curation: Ji Yang, Shuning Zhang.

Software: Ji Yang.

Writing – original draft: Ji Yang.

Writing – review & editing: Jian Liu, Lin Xin, Shijian Cao.