EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Supplementum 301: Abstracts of the annual meeting of the Swiss Society of Intensive Care Medicine (Basel, Switzerland, September 23-25, 2026)

Swiss Society Of Intensive Care Medicine

Swiss Medical Weekly · 18.08.2026 · 156 (08) · 301

DOI: 10.57187/5718 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Swiss Medical Weekly, online erschienen am 18.08.2026. Verfasst von Swiss Society Of Intensive Care Medicine. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Introduction

Diagnostic tools and scoring systems that are not specifically validated in critically ill patients often show reduced performance in the intensive care unit (ICU).This limitation also applies to tests for detecting heparin-induced thrombocytopenia (HIT), a rare immune-mediated reaction causing paradoxical thrombosis that is frequently over-suspected and difficult to diagnose accurately.We therefore aimed to evaluate the diagnostic performance of the TORADI-HIT algorithm, a machine-learning tool integrating clinical and laboratory data, in critically ill patients.

Methods

We conducted a secondary analysis and included mixed surgical-medical ICU patients from the prospective multicenter TORADI-HIT cohort across eight Swiss centers.Patients with suspected HIT were identified when either an antiheparin/PF4 immunoassay was ordered or a hematology consultation was requested.Clinical and laboratory data were collected and multiple anti-heparin/PF4 immunoassays were performed.HIT was confirmed by a positive heparin-induced platelet activation (HIPA) assay, which is considered as the reference gold standard.We compared the diagnostic performance of the TORADI-HIT algorithm with current diagnostic approaches, including the American Society of Hematology (ASH) algorithm and evaluated clinical outcomes.

Results

Among 519 ICU patients with suspected HIT, 473 (91.1%) had complete data and were included in the analysis.HIT was confirmed in n = 37 patients (7.8%).The median age was 66 years and 62.6% were male.Sepsis was present in n = 327 patients (69.1%).n = 73 patients (15.4%) had an intraarterial device, and n = 96 (20.3%) had undergone major surgery.The area under the receiver operating characteristics curve of the TORADI-HIT algorithm for presence of HIT was 0.97 (95% CI: 0.94, 1.00).The false negative rate was 8.1% (n = 3) with a false-positive rate of 6.9% (n = 30).In comparison, the ASH algorithm showed a false negative rate of 13.5% (n = 5) and a false positive rate of 9.2% (n = 40).Patients with HIT developed venous thrombosis significantly more often than those without HIT (36.1% vs. 9.4%, p < 0.001).Discussion / Conclusion: In critically ill patients, the TORADI-HIT algorithm demonstrated excellent diagnostic accuracy and offers a rapid, user-friendly, and pragmatic approach for identifying HIT.Its use may help to reduce both diagnostic delays and unnecessary treatment in the ICU.Conflicts of interest: JW and JCS are affiliated with the Department of Intensive Care, University Hospital, Inselspital which reports grants (full departmental disclosure) from Anandic Medical Systems, CSL Behring, Phagenesis, and Hemotune outside the project.

Zitieren

APA

Swiss Society Of Intensive Care Medicine (2026). Supplementum 301: Abstracts of the annual meeting of the Swiss Society of Intensive Care Medicine (Basel, Switzerland, September 23-25, 2026). Swiss Medical Weekly, 156(08), 301. https://doi.org/10.57187/5718

Vancouver

Swiss Society Of Intensive Care Medicine. Supplementum 301: Abstracts of the annual meeting of the Swiss Society of Intensive Care Medicine (Basel, Switzerland, September 23-25, 2026). Swiss Medical Weekly. 2026;156(08):301. doi:10.57187/5718

BibTeX
@article{SwissSocietyOfIntensiveCareMedicine2026,
  author = {Swiss Society Of Intensive Care Medicine},
  title = {Supplementum 301: Abstracts of the annual meeting of the Swiss Society of Intensive Care Medicine (Basel, Switzerland, September 23-25, 2026)},
  journal = {Swiss Medical Weekly},
  year = {2026},
  volume = {156},
  number = {08},
  pages = {301},
  doi = {10.57187/5718}
}
Kostenlos starten →