Ten-year incidence and heart failure dominance of hospitalized cardiovascular events in rheumatoid arthritis: a hospital-based cohort study from Martinique

Abstract

Higher risks of cardiovascular mortality and cardiovascular events have been reported in patients with rheumatoid arthritis (RA) compared with the general population. However, data remain scarce in populations of African descent, and the Caribbean region is under-represented in this field. We aimed to assess the burden of hospitalized cardiovascular events (HCE) in Martinique, a French Caribbean overseas territory with a predominantly Afro-Caribbean population. We conducted a retrospective cohort study of patients with RA from Martinique identified through the hospital discharge database. The 10-year cumulative incidence of HCE was estimated using the Fine and Gray competing-risk method, with a sensitivity analysis restricted to the classical 3-point major adverse cardiovascular events (MACE) definition. Among 205 patients with RA, 50.5% had hypertension, 30.0% diabetes mellitus,39.4% had dyslipidemia and 59.5% were overweight or obese. Smoking exposure was uncommon. 10-year cumulative incidence of HCE was 8.3% (95% CI 4.5–12.1) for the overall composite endpoint and 4.0% (95% CI 1.3–6.8) when restricted to 3-point MACE. Acute heart failure (HF) accounted for 47.1% (8/17) of events and was mostly incident, with non-ischemic etiologies. Patients with RA who developed HCE were significantly older at RA diagnosis, more frequently diabetic, and exposed to fewer bDMARDs or tsDMARDs. HCE affected a substantial proportion of patients with RA in Martinique with incidence estimates comparable to other RA cohorts despite a distinct cardiometabolic profile. HF emerged as the leading cardiovascular outcome, supporting the integration of heart failure prevention and early detection into cardiovascular surveillance strategies for RA in this setting.

Highlights

In this first study from the Caribbean region, the 10-year cumulative incidence of hospitalized cardiovascular events was 8.3%, consistent with international RA cohorts, despite a distinctive cardiometabolic profile marked by high rates of hypertension and diabetes alongside very low smoking exposure.

In this first study from the Caribbean region, the 10-year cumulative incidence of hospitalized cardiovascular events was 8.3%, consistent with international RA cohorts, despite a distinctive cardiometabolic profile marked by high rates of hypertension and diabetes alongside very low smoking exposure.

Heart failure was the leading event subtype (47.1%), and medical record review indicated heterogeneous and non ischemic etiologies, consistent with the broader non coronary cardiovascular epidemiology of Martinique.

These findings suggest that cardiovascular surveillance in Afro-Caribbean patients with RA should prioritise prevention and early detection of heart failure in addition to coronary risk reduction, with particular attention to hypertension, diabetes and non-ischemic cardiomyopathies, including conditions such as transthyretin amyloidosis where clinically suspected.

Introduction

Rheumatoid arthritis (RA) is an autoimmune disease associated with increased cardiovascular morbidity and mortality. Large cohort studies and meta-analyses have reported higher risks of cardiovascular mortality, heart failure (HF) and cardiovascular events in patients with RA than in the general population [1–3]. However, data from the literature suggest that better control of systemic inflammation with Disease-Modifying Anti-Rheumatic Drugs (DMARDs) may lower the risk of cardiovascular morbidity and mortality in RA [4–7].

Yet, evidence on the cardiovascular burden of RA in populations of African descent and from Caribbean or low-resource settings remains limited. In 2020, a systematic review highlighted the paucity of data on cardiovascular disease outcomes in racially and ethnically diverse RA populations in the United States [8]. Outside the United States, data from Afro-descendant [9, 10] or Caribbean patients with RA are scarce, and traditional cardiovascular risk factors have been reported to be highly prevalent in RA cohorts enriched in Black or Afro-descendant patients [11]. Furthermore, studies conducted in the United States have reported disparities in access to csDMARDs and bDMARDs among minority populations, with less frequent treatment escalation [12–14], a pattern that may not apply uniformly across healthcare systems. Addressing this gap is not only a scientific priority but also a question of health equity. Cardiovascular risk stratification tools built largely on European and North American data [15] may not fully translate to Caribbean and other under‑represented populations, with important implications for cardiovascular risk assessment and management in clinical practice.

Martinique is a French Caribbean territory with a predominantly Afro-Caribbean population [16, 17]. Previous local data from the EPPPRA study reported a crude prevalence of RA of 0.18% in a 297,448 adult population lower than estimates reported in European populations (0.3–0.8%) [18]. This lower prevalence may reflect the very low tobacco exposure in the Martinican population [19], a major environmental trigger for ACPA-positive RA, as well as lower genetic susceptibility to RA in populations of African ancestry, which has been associated with the degree of European genetic admixture through HLA-DRB1 shared epitope effects [20]. Martinican patients with RA also exhibit a distinctive cardiovascular profile, with low smoking exposure and substantial cardiometabolic risk factors [18]. In that study, more than 90% of patients self-identified as Afro-Caribbean [18]. At the population level, data from Santé Publique France indicate that hypertension affects 31.5% of the general adult population of Martinique versus 21.9% in mainland France, and that the prevalence of known diabetes is 11.5% versus 5.7% respectively [19, 21] .

In contrast to settings where treatment access may be limited by insurance or socioeconomic barriers, the French universal healthcare system in Martinique provides reimbursement for conventional synthetic (cs), biologic (b) and targeted synthetic (ts) DMARDs in patients with RA. Finally, the University Hospital of Martinique (CHUM) plays a central role in the care of patients with rheumatoid arthritis. With 9 of the 13 rheumatologists on the island hospital-based at the time of the EPPPRA study [18], the CHUM routinely follows patients with RA across the full spectrum of disease severity. As a result, the CHUM fulfils a primary and secondary care function for rheumatology, in addition to its role as a tertiary referral center.

Thus, Martinique is a relevant setting to study cardiovascular outcomes in RA in a usually underrepresented population. In this context, we aimed to estimate the 10-year cumulative incidence of hospitalized cardiovascular events (HCE) and assess HCE phenotype in Martinican patients with RA.

Methods

Study design and setting

We conducted a retrospective hospital-based cohort study at the CHUM, French West Indies. Acute cardiovascular care is centralized at the CHUM, and no private facility is authorized to provide inpatient care for acute cardiovascular events on the island.

Study population

Patients with a potential diagnosis of RA were identified from the French hospital discharge database, (Programme de médicalisation des systèmes d’information, PMSI), using ICD-10 codes M05 and M06. Code M06.1, corresponding to adult-onset Still’s disease, was excluded from the search algorithm. All medical records identified through PMSI screening were reviewed by two rheumatologists to confirm the diagnosis of RA, and disagreements were resolved by consensus. RA diagnosis was confirmed by two independent reviewers through comprehensive review of the complete medical records, taking into account clinical, biological, therapeutic and disease course data.

Patients were included if they had a specialist-confirmed diagnosis of RA, were aged 18 years or older at RA diagnosis, had been diagnosed between January 1, 2005, and December 31, 2015, and resided in Martinique. Patients with insufficient information to confirm the diagnosis of RA or to determine the date of RA diagnosis were excluded.

Data collection

Demographic, clinical, serological, radiological, and treatment data were extracted from electronic and available paper health records. Collected variables included age and date at RA diagnosis, sex, smoking status, rheumatoid factor (RF) and anti-citrullinated peptide antibody (ACPA) positivity, radiographic or ultrasonographic erosions, exposure to corticosteroids, csDMARDs, bDMARDs or tsDMARDs, date of last follow-up at the CHUM, and date of death when applicable.

Cardiovascular risk factors were also collected and included hypertension, diabetes mellitus, dyslipidemia, and body mass index (BMI). Hypertension, diabetes, and dyslipidemia were defined by a documented diagnosis, corresponding treatment, or biological/clinical values meeting standard diagnostic thresholds. Pre-existing cardiovascular conditions documented in the medical records prior to the index hospitalized cardiovascular event were also collected, including cardiac arrhythmias and conduction disorders, valvular heart disease, coronary artery disease, peripheral arterial disease, and dilated cardiomyopathy. Patients who had experienced a hospitalized cardiovascular event as defined by the study primary outcome prior to RA diagnosis were excluded from the analysis.

Identification of hospitalized cardiovascular events

Hospitalized cardiovascular events occurring between January 1, 2005, and December 31, 2025, were identified using a two-step approach. First, PMSI data were screened using ICD-10 codes for stroke, myocardial infarction (MI), acute coronary syndromes without MI (unstable angina) and acute heart failure (HF): I60–I64, I20–I22, I50, I130, I132, I110, and I249. After removal of duplicates, the list of patients with cardiovascular events was matched with the confirmed RA cohort.

To limit under-ascertainment related to discharge coding in PMSI, the complete electronic health records of all included patients, including rheumatology and non-rheumatology records, were systematically reviewed by two independent reviewers, and disagreements were resolved by consensus.

Cardiovascular death was identified by clinical adjudication based on medical record review, including the circumstances of death, the clinical context of the final hospitalization or last medical consultation, and any available discharge summary or death report. Deaths were identified through review of hospital records and by linkage with the French national death registry using the matchID tool (deces.matchid.io), a free, open-source search engine that exploits the Institut national de la statistique et des études économiques (INSEE) open data file of all deaths recorded in France since 1970. National death certificate data (CePIDC) were not available.

For patients who experienced hospitalized acute HF, a systematic review of the complete medical records was additionally performed to characterize echocardiographic findings, left ventricular ejection fraction, heart failure phenotype, etiology, pre-existing cardiovascular conditions and renal function.

Outcomes

The primary outcome was the first hospitalized cardiovascular event occurring within 10 years after RA diagnosis. The main composite endpoint included hospitalized acute HF, stroke, myocardial infarction (MI) or acute coronary syndromes without MI (unstable angina), and cardiovascular death.

A sensitivity analysis was performed using a restrictive 3-point MACE definition, including stroke, non-fatal MI, and cardiovascular death. Deaths occurring before a hospitalized cardiovascular event were treated as competing events.

Statistical analysis

Categorical variables were expressed as numbers and percentages, and continuous variables as mean and standard deviation, as appropriate. The 10-year cumulative incidence of hospitalized cardiovascular events was estimated using the Fine and Gray competing-risk method, with death before cardiovascular hospitalization treated as a competing event. The same method was used for the sensitivity analysis restricted to the classical 3-point MACE definition. Incidence rates were calculated per 1,000 person-years with 95% confidence intervals.

An exploratory univariate comparison was performed between patients who experienced a hospitalized cardiovascular event and those who remained event-free. Categorical variables were compared using the chi-square test or Fisher’s exact test, and continuous variables using the Mann–Whitney test, as appropriate. A Bonferroni correction was applied to account for the simultaneous comparisons performed.

All analyses were performed using R statistical Software, version 4.5.2, R Core Team 2025 [22].

Ethics

This study was approved by the Institutional Review Board of the University Hospital of Martinique (Institutional Review Board du Centre Hospitalier Universitaire de Martinique) under approval number 2023/029 on June 29th 2023. Written informed consent was not required for this retrospective study. In accordance with the General Data Protection Regulation, all eligible patients received written information by mail regarding the use of their medical data for research purposes and were informed of their right to refuse by contacting the Data Protection Officer of the University Hospital of Martinique.

Results

Characteristics of the study population

Among the 4,750 hospital stays initially identified in the PMSI database, 917 stays related to RA were retained for detailed chart review after removal of duplicates. A total of 205 patients were included in the study. (Fig. 1).

Fig. 1.

Flow chart of patients inclusion. AID: autoimmune disease; CTD: connective tissue disease; RA: rheumatoid arthritis. Non-inflammatory polyalgic syndromes included fibromyalgia and post-chikungunya arthralgia. Alternative causes of polyarthritis included gout, calcium pyrophosphate deposition disease, spondyloarthritis and septic arthritis

The demographic and clinical characteristics of the 205 patients are summarized in Table 1.

Table 1.

Characteristics of Martinican patients with RA

Variable	Patients analyzed (N)	Value
Demographic characteristics
Age at RA diagnosis, years	205	56.9 ± 14.9
Male sex	205	32 (15.6)
RA phenotype
ACPA positivity	188	103 (54.8)
RF positivity	187	100 (53.5)
Erosive disease	189	72 (38.1)
Cardiovascular risk factors
Body mass index, kg/m²	163	27.5 ± 7.1
Overweight	163	47 (28.8)
Obesity	163	50 (30.7)
Smoking	198	10 (5.1)
Hypertension	200	101 (50.5)
Diabetes mellitus	200	60 (30.0)
Dyslipidemia	170	67 (39.4)
≥ 1 cardiovascular risk factor	202	165 (81.7)
Pre-existing CVD	205	20 (9.8)
Cardiac arrhythmias and conduction disorders	205	13 (6.3)
Specifically atrial fibrillation	205	8 (3.9)
Valvular heart disease	205	6 (2.9)
Peripheral arterial disease	205	3 (1.5)
Dilated cardiomyopathy	205	1 (0.5)
Coronary artery disease	205	5 (2.4)
Treatments
Any corticosteroid use	126	97 (77.0)
Corticosteroid use for > 1 year	117	69 (59.0)
≥ 1 csDMARD	204	195 (95.6)
≥ 1 bDMARDs or tsDMARDs	201	118 (58.7)
≥ 1 tsDMARDs	201	23 (11.4)
Number of bDMARD/ tsDMARDs	205	1.1 ± 1.3

ACPA: Anti-citrullinated peptide antibody bDMARDs: biologic disease-modifying anti-rheumatic drugs; csDMARD : conventional synthetic disease-modifying anti-rheumatic drugs; tsDMARDs: target synthetic disease-modifying anti-rheumatic drugs. Overweight : BMI ≥ 25 kg/m², Obesity : BMI ≥ 30 kg/m²,  RF : Rheumatoid Factor

The mean age at RA diagnosis was 56.9 ± 14.9 years, and 84.4% of patients were female.

Regarding RA phenotype, 54.8% of patients were positive for ACPA, 53.5% for RF, and 38.1% had erosive disease.

Hypertension was present in 50.5% of patients, diabetes mellitus in 30.0%, dyslipidemia in 39.4%, and overweight and obesity in 59.5% of patients when available. Smoking exposure was uncommon (5.1%), and 81.7% of patients had at least one cardiovascular risk factor. Twenty patients (9.8%) had at least one pre-existing CVD prior to their index event. These included cardiac arrhythmias and conduction disorders in 13 patients (6.3%), coronary artery disease in 5 patients (2.4%), valvular heart disease in 6 patients (2.9%), peripheral arterial disease in 3 patients (1.5%), and dilated cardiomyopathy in 1 patient (0.5%).

Treatment exposure was substantial. Overall, 95.6% of patients had received at least one csDMARD, and 58.7% had been exposed to at least one bDMARD or tsDMARD.

Incidence of hospitalized cardiovascular events

A total of 17 hospitalized cardiovascular events were observed in RA patients.

The 10-year cumulative incidence of hospitalized cardiovascular events, estimated with the Fine and Gray competing-risk model, was 8.3% (95% CI 4.5–12.1) (Fig. 2).

When the outcome was restricted to a 3-point MACE definition including stroke, non-fatal MI, and cardiovascular death, the 10-year cumulative incidence was 4.0% (95% CI 1.3–6.8) (Fig. 2).

Fig. 2.

Ten-year cumulative incidence of hospitalized cardiovascular events: overall versus 3‑point MACE definition. The two curves correspond to different endpoint definitions applied to the same cohort. The difference between them reflects the contribution of hospitalized heart failure and acute coronary syndrome to the broader composite endpoint

The overall incidence rate in the main analysis was 8.85 events per 1,000 person-years (95% CI 5.16–14.17), compared with 4.17 events per 1,000 person-years (95% CI 1.80–8.21) for the 3-point MACE analysis.

The cumulative incidence curves showed a progressive increase in cardiovascular events over time after RA diagnosis, with most events occurring after the first 3 to 5 years of disease evolution.

HF was the leading event subtype, accounting for 8 (47.1%) of all hospitalized cardiovascular events, followed by stroke in 5 cases, myocardial infarction in 2 cases, unstable angina in 1 case, and cardiovascular death in 1 case.

Among the 17 patients who experienced a hospitalized cardiovascular event, 6 (35.3%) had at least one pre-existing cardiovascular condition prior to their index event. Specifically, 2 of 5 patients with stroke had pre-existing cardiovascular conditions (atrial fibrillation with valvular heart disease in one case, and peripheral arterial disease in the other) and the patient who experienced cardiovascular death had known coronary artery disease.

Among the 8 patients who experienced HF, results of the medical record review are summarized in Table 2. No ischemic etiology was identified in any case. HFpEF was documented in 3 patients, HFmrEF (LVEF 40–49%) in 1, and HFrEF (LVEF < 40%) in 4. Etiologies included confirmed transthyretin amyloid cardiomyopathy (ATTR)(n = 2, 25.0%), hypertensive cardiomyopathy (n = 1, 12.5%), restrictive cardiomyopathy with pulmonary hypertension (n = 2, 25.0%), rheumatic valvular disease with atrial fibrillation (n = 1, 12.5%), hypertrophic cardiomyopathy with systolic dysfunction and intracardiac thrombi (n = 1, 12.5%), and likely drug-induced dilated cardiomyopathy (n = 1, 12.5%). Three patients had pre-existing cardiovascular conditions potentially contributing to cardiac decompensation. The remaining 5 had no pre-existing CVD, suggesting incident HF in most cases. Renal function at the time of the index HF hospitalization was unavailable for 50% of patients, and follow-up values were used in those cases. No available eGFR value was below 60 mL/min/1.73 m² .

Table 2.

Echocardiographic characteristics, pre-existing cardiovascular conditions, and etiological classification of heart failure events

Year of event	Preexisting CVD	eGFR (mL/min/1.73 m²)	Echocardiographic data for heart failure during cardiovascular event or at follow up	LVEF	HF phenotype	Etiology
2009	None	eGFR 97 mL/min/1.73 m² (measured 2016, post-event)	Severe diastolic dysfunction. Follow-up TTE (02/2011): LVEF 81%, concentric LVH, diastolic dysfunction.	81% (follow-up TTE 2011)	HFpEF	Hypertensive cardiomyopathy (incident HF)
2009	Atrial fibrillation (related to hyperthyroidism)	eGFR 108 mL/min/1.73 m² (measured 2014, post-event)	Follow-up TTE (2013): LVEF 60%, biatrial dilatation, dilated non-compliant IVC, severe PH (sPAP 90 mmHg).	60% (follow-up TTE 2013)	HFpEF	Restrictive cardiomyopathy with preserved LVEF and PH (possible cardiac decompensation)
2012	Rheumatic mitral regurgitation	eGFR 80 mL/min/1.73 m² (measured 2019, post-event)	Severe mitral regurgitation with persistent atrial fibrillation. Mitral valve replacement. Follow-up TTE (08/2016): LVEF 45%, dilated hypokinetic RV, pulmonary hypertension (sPAP 55 mmHg).	45% (follow-up TTE 2016)	HFmrEF	Rheumatic valvular disease + AF (possible cardiac decompensation)
2017	None	eGFR 75 mL/min/1.73 m² (at time of event)	Hypertrophic hypokinetic cardiomyopathy. TOE: severe LVEF impairment on non-dilated heart, low output, two intracardiac thrombi (left atrial appendage 7 × 11 mm; interatrial septum 18 × 10 mm). TTE: LVEF 30%, septal thrombus 24 × 8 mm.	30%	HFrEF	Hypertrophic cardiomyopathy with systolic dysfunction; undetermined etiology (incident HF)
2013	None	eGFR 75 mL/min/1.73 m² (measured 2014, post-event)	Dilated cardiomyopathy. LVEF 20%.	20%	HFrEF	Likely drug-induced (immunosuppressive) dilated cardiomyopathy (incident HF)
2022	Atrial fibrillation History of pulmonary embolism	eGFR 77 mL/min/1.73 m² (at time of event)	Restrictive cardiomyopathy. LVEF 55%, concentric LVH, severely reduced cardiac output (2.35 L/min), severe PH, major biatrial dilatation.	55%	HFpEF	Restrictive cardiomyopathy with preserved LVEF and PH (possible cardiac decompensation)
2018	None	eGFR 66 mL/min/1.73 m² (at time of event)	Global hypokinesia, LVEF 25%, elevated LV filling pressures.	25%	HFrEF	Confirmed ATTR cardiomyopathy (incident HF)
2020	None	eGFR 64 mL/min/1.73 m² (at time of event)	LVEF 25%, cardiac output 2.5 L/min, severe LVH (IVS 20 mm), fused E/A wave, severe RV dysfunction (S’ 6 cm/s, TAPSE 10 mm), dilated non-compliant IVC.	25%	HFrEF	Confirmed ATTR cardiomyopathy (incident HF)

LVEF was assessed at the time of the index event when available, or on the nearest follow-up echocardiographic examination HF phenotype was classified according to the 2023 Focused Update of the 2021 ESC Guidelines for Acute or Chronic Heart Failure [23] (HFpEF: LVEF ≥ 50%; HFmrEF: LVEF 40–49%; HFrEF: LVEF < 40%). eGFR was estimated using the CKD-EPI formula. When eGFR was not available at the time of the index event, the nearest available value obtained during follow-up is reported. ATTR: transthyretin amyloid cardiomyopathy; eGFR: estimated Glomerular Filtration Rate; HFmrEF: heart failure with mildly reduced ejection fraction; HFpEF: heart failure with preserved ejection fraction; HFrEF: heart failure with reduced ejection fraction; IVC: inferior vena cava; IVS: interventricular septum; LV: left ventricle; LVH: left ventricular hypertrophy; LVEF: left ventricular ejection fraction; RV: right ventricle; PH : pulmonary hypertension, sPAP: systolic pulmonary artery pressure; TAPSE: tricuspid annular plane systolic excursion; TTE: transthoracic echocardiography; TOE: transoesophageal echocardiography; TR: tricuspid regurgitation

Factors associated with hospitalized cardiovascular events occurrence

Comparative analyses between patients with and without hospitalized cardiovascular events are shown in Table 3.

Table 3.

Exploratory comparison between patients with and without hospitalized cardiovascular events

Variable	Total patients analyzed (N)	Event-free RA	RA with events	p	p*
Age at RA diagnosis, years	205	56.0 ± 15.0	67.6 ± 10.2	< 0.001	0.007
Male sex	205	30 (16.0)	2 (11.8)	1.000	1.000
ACPA positivity	188	98 (56.0)	5 (38.5)	0.257	1.000
RF positivity	187	96 (54.5)	4 (36.4)	0.352	1.000
Erosive disease	189	65 (37.4)	7 (46.7)	0.581	1.000
Hypertension	200	87 (47.5)	14 (82.4)	0.009	0.235
Diabetes mellitus	200	47 (25.7)	13 (76.5)	< 0.001	0.001
Smoking	198	10 (5.5)	0 (0.0)	1.000	1.000
Dyslipidemia	170	58 (37.9)	9 (52.9)	0.296	1.000
Overweight or obesity	163	89 (59.7)	8 (57.1)	1.000	1.000
≥ 1 cardiovascular risk factor	202	148 (80.0)	17 (100.0)	0.046	1.000
≥ 1 csDMARD	204	180 (95.7)	15 (93.8)	0.528	1.000
Any corticosteroid use	126	91 (77.1)	6 (75.0)	1.000	1.000
Corticosteroid use for > 1 year	117	67 (60.4)	2 (33.3)	0.226	1.000
≥ 1 bDMARD or tsDMARDs	201	114 (61.6)	4 (25.0)	0.007	0.167
Number of bDMARD/ tsDMARDs	205	1.2 ± 1.3	0.3 ± 0.6	< 0.001	< 0.001

ACPA: Anti-citrullinated peptide antibody bDMARD: biologic disease-modifying anti-rheumatic drugs; csDMARD: conventional synthetic disease-modifying anti-rheumatic drugs; tsDMARD: target synthetic disease-modifying anti-rheumatic drugs. Overweight: BMI ≥ 25 kg/m², Obesity: BMI ≥ 30 kg/m², RF: Rheumatoid Factor

* A Bonferroni correction was applied to account for the 16 simultaneous comparisons performed

In univariate analysis, patients who experienced hospitalized cardiovascular events were older at RA diagnosis (67.6 ± 10.2 vs. 56.0 ± 15.0 years, p < 0.001) and more frequently had diabetes mellitus (76.5% vs. 25.7%, p < 0.001). These two associations remained significant after Bonferroni correction for 16 comparisons (adjusted significance threshold p < 0.0031). Hypertension (82.4% vs. 47.5%, p = 0.009) and having at least one cardiovascular risk factor (100% vs. 80.0%, p = 0.046) were each more frequent, and exposure to at least one bDMARD or tsDMARD (61.6% vs. 25.0%, p = 0.007) was less frequent in patients with events, but these differences did not remain significant following Bonferroni correction. The mean number of bDMARDs/tsDMARDs was also lower in the event group (0.3 ± 0.6 vs. 1.2 ± 1.3, p < 0.001; Bonferroni‑adjusted p < 0.001). No significant association was observed with RA phenotype, smoking, dyslipidemia, or obesity.

Discussion

We describe for the first time hospitalized cardiovascular events in a cohort of RA patients from Martinique, a French Caribbean island with a predominantly Afro-descendant population. These data help to fill an important evidence gap in RA cardiovascular outcomes by bringing information from an under-represented setting, and thereby contribute to a more globally equitable evidence base for cardiovascular risk in RA.

The cardiometabolic profile of our cohort differed markedly from that of the ESPOIR inception cohort, where hypertension and diabetes were reported in only 17.1% and 3.8% of patients respectively [24], compared with 50.5% and 30.0% in our cohort. These contrasts reflect both differences in study design and the specific cardiometabolic context of Martinique, where hypertension affects 31.5% of the general adult population, nearly 1.5 times the mainland France rate, and is particularly prevalent among women [19, 21]. This high burden of cardiometabolic comorbidities is consistent with observations from cohorts enriched in Black or Afro-descendant patients with RA [9–11], although differences in study design, healthcare systems and ethnic composition preclude direct comparisons.

Conversely, tobacco exposure was very low (5.1%), consistent with the EPPPRA study and with the low smoking prevalence in the general Martinican population [18, 25], and contrasting sharply with North American and European RA cohorts [24, 26, 27]. Low smoking exposure combined with high cardiometabolic risk thus appears as a distinctive feature of this cohort and underscores the need to broaden the cardiovascular evidence base for rheumatoid arthritis beyond traditionally studied populations.

Treatment exposure was high, with 95.6% of patients receiving at least one csDMARD and 58.7% at least one bDMARD. This partly reflects the French requirement for hospital-based initiation of biologic therapy during the inclusion period [28], which may have drawn additional patients into the hospital cohort, and contrasts with reported disparities in biologic access among African American RA populations [12–14].

Over 10 years, the cumulative incidence of hospitalized cardiovascular events was 8.3%, corresponding to an incidence rate of 8.85 per 1,000 person-years. When we restricted the outcome to a classical 3-point MACE (stroke, non-fatal MI and cardiovascular death), the 10-year cumulative incidence fell to 4.0%, with an incidence rate of 4.17 per 1,000 person-years. This gap illustrates how the choice of endpoint definition can change the estimated cardiovascular burden in RA and why comparisons across studies should be made cautiously, given the heterogeneity in design, event ascertainment and MACE definitions [29] as seen for example in cohorts such as ESPOIR and the Turkish BioSTAR registry [24, 30].

HF was the leading event subtype, accounting for 8 of 17 events (47.1%). The distribution of event subtypes also mirrors the cardiovascular epidemiology of Martinique, where age- and sex-standardized annual hospitalization rates in the general adult population are 299 per 100,000 for HF, 295 per 100,000 for stroke, and 235 per 100,000 for ischemic heart disease, contrasting with mainland France, where ischemic heart disease is the leading cause of cardiovascular hospitalization (459 per 100,000) [31]. Although these population-level figures are not directly comparable to our cohort, they suggest that the predominance of HF and stroke in our RA patients is consistent with, and may partly reflect, the broader non-coronary cardiovascular profile of Martinique.

RA has been linked to an increased risk of HF [32, 33], and the high frequencies of hypertension and diabetes in our cohort are consistent with established HF risk factors in RA [34–36]. Most events instead reflected heterogeneous, non‑ischemic etiologies, although a contributory role of the underlying cardiometabolic burden (hypertension, diabetes) cannot be excluded in individual cases. ATTR was confirmed in 2 patients, consistent with the known prevalence of Val122Ile ATTR cardiomyopathy in Afro-Caribbean populations and illustrating the contribution of infiltrative, non-ischemic cardiomyopathies to HF in this setting [37]. While current EULAR recommendations for cardiovascular risk management in RA focus primarily on atherosclerotic risk factor identification and validated ischemic risk scores [15], and their implementation in routine practice is often suboptimal [38], our data suggest that cardiovascular surveillance in Martinican patients with RA should prioritize HF prevention and early detection, alongside coronary risk reduction, with particular attention to hypertension, diabetes and non-ischemic cardiomyopathies when clinically suspected.

Classical cardiovascular risk factors appeared more frequent among patients who developed hospitalized cardiovascular events: they were older at RA diagnosis, more frequently diabetic and had fewer bDMARDs or tsDMARDs exposure after Bonferroni correction. In contrast, RA-related characteristics (ACPA, RF, erosive disease) did not differ significantly between patients with and without events.

The absence of a significant association between ACPA or RF positivity and hospitalized cardiovascular events in our cohort warrants careful interpretation. Several studies have suggested that ACPA-positive RA may carry greater cardiovascular risk, possibly mediated by higher inflammatory burden and more severe disease phenotype [39]. However, multiple factors may explain the lack of association in our cohort. The limited number of events substantially reduced statistical power to detect associations with serological variables, particularly given that ACPA and RF data were missing in a subset of patients. Moreover, serological status was recorded as a binary variable (positive/negative) rather than as antibody titres, which may have obscured a potential dose–response relationship. Finally, longitudinal measures of RA disease activity and cumulative inflammatory burden, which have been implicated as important contributors to cardiovascular risk in RA and may partly explain the associations observed between RA immunological phenotype and cardiovascular outcomes in previous studies, were not systematically available in this retrospective cohort [40]. The composite endpoint in our study was dominated by HF (47.1% of events) rather than atherothrombotic events. Most studies reporting an association between ACPA positivity and cardiovascular risk have focused on atherosclerotic outcomes, such as ischemic heart disease and stroke. In contrast, the relationship between ACPA/RF and non-ischemic HF phenotypes, which predominated in our cohort, has been far less explored and is not clearly established. Taken together with the small number of events and missing serological data, these considerations preclude firm conclusions regarding the role of RA immunological phenotype in cardiovascular risk in this population.

The lower mean number of bDMARDs/tsDMARDs in patients who experienced cardiovascular events should be interpreted with caution. The small number of events precluded multivariable analysis, and our comparisons should be read as descriptive rather than as evidence of independent associations.

Our study has several strengths. To our knowledge, it is the first to describe hospitalized cardiovascular outcomes in an RA cohort from the Caribbean region, in a French Caribbean setting with a predominantly Afro-descendant population. The long follow-up, the use of a competing-risk approach, and the dual strategy for outcome ascertainment all strengthen the analysis. Acute cardiovascular care in Martinique is highly centralized: the University Hospital of Martinique is the main referral center for acute cardiovascular events, and no private facility is authorized to manage acute cardiovascular emergencies. This specific organization likely improved the completeness of cardiovascular event capture. In addition, PMSI discharge data were complemented by systematic review of full electronic medical records by two independent reviewers, allowing detection of cardiovascular events that might not have been recorded in discharge codes.

Several limitations need to be acknowledged. A major limitation is the absence of an age- and sex-matched control cohort from the general Martinique population, which precludes formal estimation of the excess cardiovascular risk attributable to RA in this setting. Our study was designed as a first descriptive incidence study in an underrepresented population rather than a comparative risk study, and population-level cardiovascular hospitalization data from Santé Publique France [31] are discussed as contextual benchmarks rather than formal comparators. The retrospective single-center design introduced a risk of selection bias and led to missing data. Reliance on electronic and paper health records precluded standardized measurement of rheumatoid arthritis activity over time. Composite disease activity indices such as the DAS28 or CDAI were not consistently recorded in routine care, which prevented us from examining in detail how longitudinal inflammatory control might have influenced cardiovascular risk. Echocardiographic data and renal function were not systematically available at the time of the index HF hospitalization. In several patients, TTE results were retrieved from subsequent assessments, which may not accurately reflect the hemodynamic status at acute presentation. As a consequence, the classification of HF events as incident or possible cardiac decompensation, as well as the assessment of left ventricular ejection fraction, should be interpreted with caution.

The organization of rheumatology care in Martinique also shapes how our results should be interpreted. Because the CHUM follows many RA patients in routine care, including those without severe disease, our cohort is probably less skewed towards the most complex cases than hospital‑based registries from strictly tertiary centres. This likely limits, but does not remove, the risk of overestimating cardiovascular event rates due to enrichment in high-risk profiles. Patients managed exclusively in private outpatient settings, particularly those with milder disease and fewer comorbidities, may still be under‑represented, so our estimates should be viewed as hospital‑based rather than fully population‑based.

Although we used PMSI data to screen for events, misclassification is still possible because of coding errors or lack of clinical detail, and non-hospitalized events, events occurring outside Martinique, or events managed without admission may have been missed. Two out-of-hospital deaths were treated as competing events rather than cardiovascular events because both patients had recently been hospitalized for non-cardiovascular conditions (RA-associated interstitial lung disease in one case and pulmonary infection with worsening pulmonary graft-versus-host disease after leukemia treatment in the other). In the absence of national death certificate data, this classification cannot be formally confirmed, so some uncertainty remains. Finally, individual-level ethnicity was not collected, in line with French regulations, but the predominantly Afro-Caribbean nature of the cohort is supported by the demographic profile of Martinique and by the EPPPRA study [18], in which more than 90% of RA patients self-identified as Afro-Caribbean.

In conclusion, over a 10-year period, the cumulative incidence of hospitalized cardiovascular events was 8.3% in this Afro-Caribbean RA cohort, corresponding to an incidence rate of 8.85 events per 1,000 person-years, which is comparable to estimates reported in international RA cohorts despite the distinctive cardiometabolic profile of this population. When restricted to a classical 3-point MACE definition, the 10-year cumulative incidence was 4.0%, illustrating how endpoint definition substantially influences the estimated cardiovascular burden in RA. Acute HF was the leading event subtype, accounting for nearly half of all HCE, and medical record review suggested heterogeneous, predominantly non-ischemic etiologies, including ATTR. In exploratory analyses, HCE was associated with older age at RA diagnosis and diabetes mellitus. These findings support an integrated cardio-rheumatology approach with particular attention to early HF detection alongside coronary risk reduction and metabolic risk factor management in Afro-Caribbean patients with RA. They also contribute to ongoing efforts to address the evidence gap in Caribbean rheumatology, where epidemiological data remain scarce [41]. In this context, Martinique provides a valuable setting combining access to the French healthcare system, comprehensive hospital data, and advanced RA therapies.

Author contributions

PR: conceptualization, visualization, data curation, methodology, validation, formal analysis, writing original draft. ADL: investigation, data curation, writing original draft. PN: validation, writing review and editing. CD: validation, writing review and editing. MD: formal analysis, validation, writing review and editing. BS: data curation, validation, writing review and editing. FLS:. conceptualization, visualization, investigation, data curation, methodology, validation, writing review and editing.

Funding

This work received no funding.

Declarations

Conflict of interest

We declare no competing interests.

AI generative statement

No part of this report is copied, published elsewhere, or generated by AI.