Dynamic Co-Evolution of C-Reactive Protein and Albumin in Early Sepsis: A Group-Based Dual-Trajectory Analysis for Risk Stratification

The coupled evolution of inflammation and metabolic dysfunction during early sepsis is incompletely characterized. Although C-reactive protein (CRP) and albumin (ALB) are established markers of these processes, their joint longitudinal patterns have rarely been examined. We used group-based dual-trajectory modelling (GBDTM) to identify distinct coevolutionary patterns of CRP and ALB and to evaluate their associations with mortality in patients with sepsis. This retrospective observational study included patients admitted to the ICU of Beijing Hospital of Traditional Chinese Medicine, Capital Medical University, between June 2014 and March 2025. All included adults met the Sepsis-3 criteria. Among 462 patients with sepsis, serial CRP and ALB measurements obtained on days 1, 3, 5, 7 and 14 after ICU admission were analysed using GBDTM. The selected three-class model showed acceptable classification quality (entropy = 0.80; class-specific average posterior probabilities [AvePPs] = 0.84-0.92) and identified slow-decreasing (79.65%), rapid-decreasing (10.39%) and increasing (9.96%) inflammatory phenotypes. The 28-day mortality rate was highest in the increasing group (52.2% vs. 32.6% and 33.3%; log-rank p = 0.034). In the fully adjusted Cox model, the increasing trajectory was associated with 28-day mortality (HR 1.943, 95% CI: 1.202-3.140; p = 0.007) and 60-day mortality (HR 1.946, 95% CI: 1.218-3.108; p = 0.005). Parsimonious models adjusted for age, sex, and SOFA or the Acute Physiology and Chronic Health Evaluation II (APACHE II) score yielded broadly similar estimates. Joint CRP-ALB trajectories identify clinically interpretable dynamic phenotypes during early sepsis. These findings support the joint trajectory as a hypothesis-generating, complementary phenotyping marker that requires external prospective validation.