# Who is at risk? Clinical features and a predictive model for 30-day mortality in hematologic patients with enterococcal bloodstream infection

## TL;DR

This study identifies risk factors and creates a predictive model for 30-day mortality in hematologic patients with enterococcal bloodstream infections.

## Contribution

A novel predictive model and risk score for 30-day mortality in hematologic patients with EBSI is developed and validated.

## Key findings

- Enterococcus faecium was the leading pathogen in EBSI cases.
- A risk score stratified patients into low- and high-risk groups with significantly different mortality rates.
- The predictive model showed good discrimination (AUROC 0.79) and calibration.

## Abstract

Enterococcal bloodstream infection (EBSI) carries high mortality in hematologic patients, yet no prognostic model tailored to this population exists.

We retrospectively analyzed 192 hematologic patients (≥14 years) with EBSI admitted between 2014 and 2024. Clinical features, microbiology, treatment, and outcomes were assessed. Candidate predictors for 30-day mortality were selected by LASSO and entered into multivariable logistic regression. A simplified risk score was derived from regression coefficients and internally validated by bootstrap resampling.

The median patient age was 43 years, and acute leukemia was the predominant underlying disease (72.4%). Enterococcus faecium was the leading pathogen (71.4%), with low vancomycin resistance (1.6%). Most cases (71.9%) occurred as breakthrough infections, mainly during carbapenem therapy, and 72.9% met mucosal barrier injury laboratory-confirmed bloodstream infection criteria. The 14- and 30-day all-cause mortality rates were 13.5% and 22.4%, respectively. Independent predictors of 30-day mortality included age ≥50 years (aOR=2.29, p=0.038), severe graft-versus-host disease (aOR=6.06, p=0.003), septic shock (aOR=30.01, p<0.001). The final predictive model, incorporating these three factors along with pneumonia and high-risk hematologic disease, demonstrated optimal discrimination (AUROC 0.79, 95% CI 0.705–0.867) and calibration. A derived risk score stratified patients into low- (<2 points) and high-risk (≥2 points) groups, with markedly different 30-day mortality (11.3% vs. 39.0%, P<0.001).

In hematologic patients, EBSIs commonly arise as breakthrough infections despite broad-spectrum antibiotic coverage, most often associated with mucosal barrier injury. Our parsimonious risk score enables early identification of patients at high risk of 30-day mortality to guide timely interventions.