# Impact of post-decannulation high fever on mortality in patients with severe ARDS treated with veno-venous ECMO: a multicenter retrospective study

## TL;DR

High fever after ECMO removal in severe ARDS patients may affect survival differently depending on infection status.

## Contribution

Identifies post-decannulation high fever as a variable with context-dependent mortality impact in ARDS patients.

## Key findings

- 23.2% of patients developed high fever within 3 days after ECMO decannulation.
- High fever was linked to reduced mortality in infected patients but increased mortality in non-infected patients.
- Overall, high fever was not significantly associated with 90-day mortality in the full cohort.

## Abstract

Few studies have examined the prognostic impact of high fever after decannulation from veno-venous (V-V) extracorporeal membrane oxygenation (ECMO) in patients with severe acute respiratory distress syndrome (ARDS). We aimed to investigate the incidence and prognostic significance of post-decannulation high fever in this population, exploring its association with mortality, stratified by the presence of infectious complications at decannulation.

This study was a post hoc analysis of a multicenter retrospective registry that included adult patients with severe ARDS who were successfully weaned off V-V ECMO between 2012 and 2022 across 24 institutions in Japan. High fever was defined as a core body temperature of ≥ 39.0 °C within 3 days after ECMO decannulation. The primary outcome was 90-day in-hospital mortality. Cox proportional hazards models were used to examine the association between post-decannulation high fever and mortality. As a subgroup analysis, we evaluated this association according to the presence or absence of infectious complications.

Among 522 patients, 121 (23.2%) developed high fever within 3 days after ECMO decannulation. In the overall cohort, 90-day in-hospital mortality did not differ significantly between the high-fever and no-fever groups (19.0% vs. 13.7%, p = 0.372). Multivariable analysis showed no statistically significant association between high fever and mortality (hazard ratio [HR] 0.92, 95% confidence interval [CI] 0.55–1.56, p = 0.770). Subgroup analyses revealed opposite associations depending on infection status. High fever was associated with reduced mortality in patients with infection (HR 0.33, 95% CI 0.12–0.89, p = 0.045) but increased mortality in those without (HR 2.25, 95% CI 1.23–4.11, p = 0.011).

Post-decannulation high fever occurs in nearly one-fourth of patients with severe ARDS treated with V-V ECMO. Its association with mortality appears to differ depending on the infection status at decannulation, underscoring the importance of carefully assessing infectious complications.