Abstract
BACKGROUND: The hemoglobin, albumin, lymphocyte, and platelet (HALP) score represents a meaningful predictor in many cardiovascular diseases. However, the predictive utility of this score for the outcome of patients admitted to the intensive care unit (ICU) due to acute myocardial infarction (AMI) has yet to be fully elucidated. METHODS: Information from the Medical Information Mart for Intensive Care (MIMIC)-IV v3.1 database was used to analyze the association between the HALP score and 90 days and 365 days all-cause mortality in critically ill patients with AMI. Patients were grouped according to the calculated HALP quartiles. Cox proportional hazards regression analysis and restricted cubic spline (RCS) analysis were performed to assess the association between the HALP score and mortality risk. A recursive algorithm identified the HALP inflection point, thus defining high and low HALP groups for the Kaplan-Meier survival analysis. Subgroup analyses analyzed the robustness across clinical strata. Furthermore, predictive models based on machine learning algorithms that included the HALP score were constructed to estimate 90 days mortality. The performance of the models was evaluated using the area under the receiver operating characteristic curve (AUC). RESULTS: A total of 818 AMI patients were included. The analysis revealed mortality rates of 31% at 90 days and 40% at 365 days. Elevated HALP values were independently linked to a reduced risk of death. In fully adjusted models, patients in the top HALP quartile exhibited significantly lower all-cause mortality at 90 days (hazard ratio (HR) = 0.68; 95% confidence interval (CI): 0.47-0.99; p = 0.047) and 365 days (HR = 0.66; 95% CI: 0.47-0.90; p = 0.011). A nonlinear, inverse "L-shaped" association was observed, with an inflection point identified at a HALP value of 19.41. Below this value, each unit increase in the HALP score reduced mortality risk by 2.4%-2.7%. The Kaplan-Meier curves confirmed an improved survival above the threshold. Meanwhile, the subgroup analyses revealed a generally consistent association between the HALP score and mortality, except for age, where a significant interaction was observed (p = 0.003), indicating a stronger protective effect in older patients. Machine learning analyses supported the robustness and predictive value of the HALP score, with a maximum AUC of 0.7804. CONCLUSIONS: The HALP score is significantly associated with all-cause mortality among critically ill individuals suffering from AMI.