# Prediction Scores Identifying Patients at High Risk of Endocarditis in Enterococcal Bacteremia

## TL;DR

This study evaluates and improves prediction scores to identify patients with enterococcal bacteremia at high risk of endocarditis, aiming to reduce unnecessary imaging.

## Contribution

The study introduces a modified DENOVi score that excludes subjective criteria and includes intracardiac devices, offering a more balanced diagnostic approach.

## Key findings

- The DENOVi score classified 42% of patients as high risk for endocarditis with a low negative likelihood ratio of 0.04.
- Adapted NOVA and DENOVi scores outperformed DENOVA in diagnostic safety and resource efficiency.
- DENOVi reduced the proportion of required echocardiograms compared to the adapted NOVA score.

## Abstract

Clinical prediction scores such as NOVA and DENOVA aim to identify patients with enterococcal bacteremia at low risk of infective endocarditis (IE) in whom imaging might be safely avoided. The aim was to evaluate the performance of NOVA and DENOVA scores and to introduce a modified tool, DENOVi.

This retrospective study included adult patients with enterococcal bacteremia at 2 Swiss tertiary centers (2015–2024). IE was adjudicated by multidisciplinary Endocarditis Teams according to 2023 Duke-International Society of Cardiovascular Infectious Diseases criteria. Patients were stratified as high risk for IE using the adapted NOVA score (cutoff: ≥4), the DENOVA score (≥3), and a newly developed DENOVi score (≥2), which excluded the subjective murmur criterion and broadened “valve disease” to include intracardiac electronic devices (new Vi component).

Among 827 bacteremia episodes, 172 (21%) were diagnosed with IE. The adapted NOVA, DENOVA, and DENOVi scores classified 76%, 26%, and 42% of patients as high risk, respectively. Corresponding NLRs were 0.04 (95% CI, .01–.15), 0.10 (0.06–0.16), and 0.04 (0.02–0.10). The adapted NOVA substantially increased the proportion of echocardiograms needed to be performed from 58% based on clinical evaluation alone to 76%, whereas the DENOVA and DENOVi scores would have reduced this proportion to 26% and 42% of episodes, respectively.

Both adapted NOVA and DENOVi scores reliably ruled out IE, but DENOVi provided the most balanced approach between diagnostic safety and resource utilization. DENOVi therefore represents a pragmatic and objective tool for IE risk stratification in enterococcal bacteremia. Prospective validation is warranted.

Among 827 episodes of enterococcal bacteremia, the adapted NOVA and DENOVi scores demonstrated low negative likelihood ratios (0.04), compared with DENOVA (0.10). The adapted NOVA, DENOVA, and DENOVi scores classified 76%, 26%, and 42% of patients as high risk, respectively.