Developing and validating a lactate-to-albumin ratio-enhanced score for mortality prediction in ICU patients with acute pulmonary embolism: a multi-cohort study

Background The simplified Pulmonary Embolism Severity Index (sPESI) has limited predictive accuracy for critically ill patients with acute pulmonary embolism (APE) and multi-organ dysfunction and lacks dynamic pathophysiological indicators. This study evaluated the predictive value of the lactate-to-albumin ratio (LAR) for 28-day mortality in ICU patients with APE and explored its potential to optimize the sPESI score. Methods This study retrospectively analyzed 1098 adult patients with APE in the ICU from the Medical Information Mart for Intensive Care-IV (MIMIC-IV) (derivation cohort, n = 306), the eICU Collaborative Research Database (eICU-CRD) (first validation cohort, n = 555), and an independent external cohort from the First Affiliated Hospital of Dalian Medical University (second validation cohort, n = 237). Multivariable Cox regression, restricted cubic spline (RCS), and Kaplan–Meier survival analyses were used to assess the association between LAR and ICU 28-day all-cause mortality. The predictive efficacy and clinical utility of LAR and the new scoring system (LAR-sPESI) were evaluated using receiver operating characteristic (ROC) curves, Hosmer–Lemeshow (H–L) tests, and decision curve analysis (DCA). Results LAR demonstrated a linear relationship with 28-day mortality. The AUC of LAR for predicting mortality was 0.610 in the derivation cohort, 0.710 in the first validation cohort, and 0.647 in the second validation cohort. The LAR-sPESI score showed modest improvements in predictive performance (derivation cohort AUC = 0.650; first validation cohort AUC = 0.733; second validation cohort AUC = 0.701), with acceptable calibration (H–L test, P > 0.05 in all cohorts). DCA suggested trends suggestive of potential clinical utility within 15–50% in the derivation cohort, 1%–80% in the first validation cohort, and 3%–80% in the second validation cohort. LAR exhibited moderate discriminative ability for norepinephrine use in all cohorts. In contrast, its predictive performance for pharmacological thrombolysis and mechanical ventilation was limited, with acceptable calibration for all assessable secondary outcomes (all P > 0.05). Conclusion LAR may serve as a preliminary exploratory predictor of 28-day mortality in ICU patients with APE. The LAR-sPESI score yielded only modest numerical gains in predictive performance, supporting exploratory risk stratification.