External validation of the EHMRG30-ST score for short-term mortality risk stratification in European ADHF patients
A Mkrtchjan, D Mouha, M C Van Herwerden, K Veen, A A Constantinescu, K Caliskan, O C Manintveld, J J Brugts, R A De Boer, L Feyz, R M A Van Der BoonAbstract
Introduction
Acute decompensated heart failure (ADHF) imposes a significant burden on healthcare systems due to high hospitalization rates, prolonged admissions, and frequent early readmissions, resulting in substantial economic and resource strain. Hospital-at-home (HaH) programs may reduce inpatient burden, but their safe implementation requires robust tools to identify suitable patients. Accurate prognostic tools can improve decision-making in ADHF by identifying high-risk patients for inpatient care and monitoring, while allowing safe early discharge for low-risk patients. However, externally validated risk stratification instruments for European ADHF populations are not available.
Purpose
To externally validate discrimination and calibration of the Emergency Heart failure Mortality Risk Grade for 30-days (EHMRG30-ST) score in a European cohort of patients hospitalized with ADHF.
Methods
We conducted a retrospective cohort study at a tertiary academic center in the Netherlands, including consecutive patients admitted with ADHF between January and December 2022. To be included, patients were required to have complete data for the calculation of the EHMRG30-ST score, which consists of demographic, clinical, and laboratory variables at the time of admission. Based on these scores, patients were subsequently classified into predefined risk strata. The primary outcome was 30-day all-cause mortality. Discrimination was assessed using receiver operating characteristic analysis, and calibration by calibration plots and Hosmer-Lemeshow test.
Results
Among 270 eligible patients, 193 had complete data for EHMRG30-ST calculation. The cohort was 60.1% male (N=116) with a median age of 71 (IQR 61-77) years. Most were classified as low risk (60.6%), with decreasing numbers in intermediate (25.4%) and high-risk (14.0%) groups. Thirty-day mortality increased across risk strata, with the highest mortality (37.0%) in the high-risk group. The area under the curve for discriminative performance was 0.749 (95% CI 0.649-0.848). Predicted probabilities did not agree with observed outcomes (Hosmer-lemeshow P-value<0.001), with general underestimation of predicted risks.
Conclusion
In a European cohort of patients hospitalized with ADHF, the EHMRG30-ST risk score demonstrated fair discriminative performance but limited calibration. While the score may support stratification of (low risk) patients who could benefit from HaH or early discharge strategies, recalibration and prospective validation are necessary before routine clinical use and individual risk communication.For image description, please refer to the figure legend and surrounding text.Figure 2 Mortality and Calibration PlotFor image description, please refer to the figure legend and surrounding text.