C-reactive protein-to-albumin ratio as a novel prognostic biomarker for long-term mortality in pericarditis: a real-world study

Pericarditis is a heterogeneous inflammatory condition with variable clinical outcomes. Although traditional inflammatory biomarkers such as C-reactive protein (CRP) and erythrocyte sedimentation rate (ESR) are routinely used for diagnosis and monitoring, they do not fully capture the interplay between inflammation, hepatic synthetic function, and nutritional status. The CRP–to–albumin ratio (CAR), a composite index integrating these components, has shown prognostic value in several cardiovascular disorders. However, its significance in pericarditis remains unknown. This was a real-world retrospective cohort study of adult patients hospitalized for pericarditis between January 1st, 2005 to December 31st, 2019 from a single tertiary centre. CAR was calculated as CRP (mg/L) divided by serum albumin (g/L) and categorized into quartiles. The primary outcome was all-cause mortality. Associations were examined using Cox proportional hazards models, restricted cubic splines (RCS), and segmented Cox regression, with prespecified sensitivity analyses using alternative adjustment strategies and modeling specifications. A total of 546 patients (mean age, 59.2 ± 16.4 years; 56.8% men) were analyzed. During a median follow-up of 64 months, 239 deaths (43.8%) occurred. Higher CAR quartiles were associated with progressively increased mortality (log-rank P < 0.001). In multivariable Cox models adjusting for demographics and comorbidities, each unit increase in CAR conferred a 5% higher mortality risk (hazard ratio [HR], 1.05; 95% confidence interval [CI], 1.01–1.10; P = 0.031). Compared with the lowest quartile, adjusted HRs for mortality were 2.04 (95% CI, 1.34–3.09), 2.21 (95% CI, 1.47–3.32), and 1.93 (95% CI, 1.26–2.95) across quartiles 2–4 (P for trend = 0.004). RCS and segmented Cox analyses demonstrated a nonlinear relationship with a threshold near CAR = 0.33—below which mortality risk increased sharply and plateaued thereafter, which should be interpreted as exploratory given limited event density at very low CAR values. The overall association remained directionally consistent in sensitivity analyses including alternative knot specifications and approaches to address extreme values. Associations were consistent across age, sex, hypertension, and malignancy subgroups. CAR independently predicted long-term all-cause mortality in patients hospitalized for pericarditis, exhibiting a nonlinear dose–response pattern. CAR represents a simple, inexpensive, and readily available biomarker that integrates inflammatory and nutritional status, with robust findings across multiple sensitivity analyses, offering incremental prognostic value beyond traditional risk factors.