Association of a dual-index phenotype and early ICU reassessment with 28-day in-hospital mortality in cardiogenic shock: a single-center observational cohort study.

BACKGROUND

Cardiogenic shock (CS) carries high short-term mortality. In this study, we test whether coexisting hypoalbuminemia and hyperchloremia (dual-index phenotype, DIP) and early intensive care unit (ICU) reassessment are associated with 28-day in-hospital mortality.

METHODS

In this single-center prospective observational cohort study (January 2021-December 2025), DIP was defined as albumin <35 g/L and chloride >110 mmol/L on paired laboratories within 0-6 h of ICU admission. We used a prespecified 30 h landmark: patients alive and still in the ICU at 30 h formed the analytic cohort, and outcome follow-up began at the landmark. A subset with a second paired assessment at 18-30 h was categorized as non-DIP → non-DIP, non-DIP → DIP, DIP → non-DIP, and DIP → DIP. Sensitivity analyses included baseline-only adjustment, propensity-score overlap weighting, Fine-Gray competing-risk modeling, and exploratory incremental prognostic performance assessment.

RESULTS

Among 982 screened admissions, 444 patients alive and still in the ICU at 30 h comprised the analytic cohort; 74/444 (16.7%) had baseline DIP. The 28-day in-hospital mortality rate was 18.2% (81/444) and was higher in DIP than in non-DIP (37.8% vs. 14.3%). In the 30-h conditional model, baseline DIP remained associated with mortality [adjusted odds ratio (aOR) 2.46, 95% confidence interval (CI) 1.30-4.64; p = 0.005]. The results were directionally consistent in a baseline-only model (aOR 2.71, 95% CI 1.45-5.06), a propensity-score overlap-weighted model (OR 2.39, 95% CI 1.27-4.50), and a Fine-Gray competing-risk model [subdistribution hazard ratio (sHR) 2.31, 95% CI 1.35-3.95]. Adding DIP to the clinical landmark model modestly improved discrimination (C-statistic 0.782-0.801). In the reassessment subset (n = 96), the mortality rate was 8.3% in non-DIP → non-DIP and 60.0% in DIP → DIP, with wide uncertainty due to small subgroup sizes.

CONCLUSIONS

Among CS patients alive and still in the ICU at 30 h, baseline DIP and persistence of DIP by approximately 24 h identified a subgroup at higher risk of 28-day in-hospital death. The association was robust in baseline-only, propensity-weighted, and competing-risk sensitivity analyses, but trajectory findings remained exploratory because of small subgroup sizes. These findings apply to the selected 30 h survivor cohort and should not be used as an ICU-admission prognostic model or extrapolated to early deaths, early ICU transfers, or patients without paired early laboratories without external validation.