The J-shaped association between the hemoglobin, albumin, lymphocyte, and platelet (HALP) score and all-cause mortality in patients with cardiometabolic multimorbidity: A retrospective cohort study

The hemoglobin, albumin, lymphocyte, platelet (HALP) score integrates nutritional, inflammatory, and immune markers. While its prognostic utility is known in other diseases, its significance in cardiometabolic multimorbidity (CMM) remains unexplored. This study aimed to investigate the association between the HALP score and all-cause mortality in a nationally representative CMM cohort. We analyzed 1479 individuals with CMM from NHANES (1999-2018). Survival and mortality risk were assessed using Kaplan-Meier methods and weighted Cox regression. Restricted cubic splines (RCS) were used to model the non-linear relationship. Exploratory time-dependent receiver operating characteristic analysis was performed on subgroups stratified by the RCS-identified inflection point to further explore this association. During a median follow-up of 68 months, survival was lowest in HALP quartile 1 (Q1). In fully adjusted models, higher quartiles (Q2-Q4) had lower mortality risk versus Q1, with hazard ratios (HRs) of 0.61, 0.65, and 0.63, respectively (all P < .01). RCS analysis revealed a J-shaped association (P for non-linearity < .001), with a nadir at 50.01. Below this nadir, HALP was inversely associated with mortality (HR per unit increase = 0.981, P < .001); above it, a small-magnitude positive association was observed (HR = 1.002, P = .018). This pattern was consistent across most subgroups, though an interaction with drinking status was noted (P for interaction = .028). The HALP score has an independent, J-shaped association with all-cause mortality in CMM patients. This highlights the complex role of the nutrition-inflammation-immunity axis. Rather than a simple prognostic tool, HALP may serve as a nuanced indicator that helps differentiate between chronic frailty and potential acute systemic failure, guiding clinical interpretation. However, these findings should be interpreted with caution, as the observational design precludes causal inference and the potential for residual confounding from unmeasured factors remains.