Non-linear association of admission platelet-to-lymphocyte ratio with 28- and 90-day mortality in critically ill patients with rheumatoid arthritis: a retrospective cohort study

Abstract

Background: Critically ill patients with rheumatoid arthritis (RA) have high short-term mortality. The platelet-to-lymphocyte ratio (PLR) reflects RA inflammatory activity. We evaluated whether admission PLR is associated with 28- and 90-day mortality in critically ill patients with RA.

Methods: RA patients were identified from the MIMIC-IV database. Restricted cubic spline (RCS) Cox models characterized the PLR-mortality association and defined a log PLR threshold. Patients were stratified into low, middle, and high PLR groups using this threshold with a ± 0.3 margin (δ). Cumulative incidence of death was assessed using the Kaplan-Meier method. Piecewise Cox models were constructed with incremental adjustment. Subgroup analyses tested the consistency of PLR's effect across different strata.

Results: Among 989 patients, 28- and 90-day mortality were 16.4% and 23.5%. RCS analysis showed a U-shaped association, with the lowest risk at log PLR = 4.725 and higher hazards toward both extremes. In piecewise Cox models, each + 1 log PLR was associated with lower mortality in the low PLR segment (28-day hazard ratio (HR) 0.59; 90-day HR 0.61) but higher risk in the high PLR segment (28-day HR 1.72; 90-day HR 1.67). Associations persisted after multivariable adjustment, with consistent subgroup trends.

Conclusions: Admission PLR is a significant prognostic marker in critically ill RA, demonstrating a U-shaped relationship with 28- and 90-day mortality. Both abnormally low and high PLR values were associated with higher mortality. Key Points • Admission PLR shows a U-shaped, independent association with 28- and 90-day mortality, with the lowest risk at log PLR 4.725 (~ PLR ≈112). • Threshold groups (PLR < ~ 84, 84 -153) and piecewise Cox models reveal decreasing risk below and increasing risk above the cutoff, persisting after multivariable adjustment. • Effects remain after multivariable adjustment and are consistent across clinical subgroups. • As a routine CBC metric, PLR enables simple early ICU risk stratification for RA, highlighting excess risk at both extremes.

Keywords: Critical illness; Inflammation; Intensive care; Mortality; Platelet-to-lymphocyte ratio; Rheumatoid arthritis.