%0 Journal Article 
%A Gutiérrez Gutiérrez, Belén
%A Gallego-Mesa, Belén
%A Kaasch, Achim J.
%A Riediger, Matthias
%A Rieg, Siegbert
%A Trigo, Marta
%A Salto-Alejandre, Sonsoles
%A Anguita-Santos, Francisco
%A Cano, Ángela
%A Prolo-Acosta, Andrea
%A López Cardenas, Salvador
%A Pérez Rodríguez, María Teresa
%A Martínez-Marcos, Francisco Javier
%A Merino-Lucas, Esperanza
%A Anaya-Baz, Blanca
%A Arizcorreta Yarza, Ana
%A Piscaglia, Marco
%A Aceituno, Alexandra
%A Romero-Calderón, Lidia
%A Hornuss, Daniel
%A Alemán-Rodríguez, Aurora
%A Gimeno-Gascón, Adelina
%A Recacha, Esther
%A Torre-Cisneros, Julián
%A Merchante, Nicolás
%A Pascual, Álvaro
%A López Cortés, Luis Eduardo
%A Rodríguez Baño, Jesús
%T Identification and validation of clinical phenotypes in Staphylococcus aureus blood stream infection and their association with mortality (FEN-AUREUS cohort-based study)
%D 2025
%@ 2589-5370
%U http://hdl.handle.net/10498/38159
%X Background: Staphylococcus aureus bacteraemia (SAB) is heterogeneous in patients and infection-related features. The aim of the study was to identify clinical phenotypes among patients with SAB, to evaluate their association with mortality, and to derive and validate a simplified probabilistic model for phenotypes assignment. Methods: Phenotypes were derived using two-stage cluster analysis of 2128 patients from the ISAC cohort (recruited between 2013 and 2015), analysing 62 variables. Cox regression assessed phenotype–mortality associations. Logistic regression was employed to develop a simplified probabilistic model for sub-phenotype allocation, validated in two external international cohorts: INSTINCT (1217 patients, recruited between 2006 and 2011) and FEN-AUREUS (1185 patients, recruited between January 2021 and October 2024). The association between sub-phenotypes and 30-day mortality in the validation cohorts was also assessed. Findings: Cluster analysis identified three clinical phenotypes based on the probable portal of entry: A (skin and soft tissues), 458 cases; B (vascular device-associated), 573 cases; and C (other portals of entry or unknown), 1097 cases. Their 30-day mortality was significantly different (13·1%, 18·2% and 25·3%, respectively, p < 0·001). Each phenotype contained two sub-phenotypes with differing characteristics and mortality risks. Also, three phenotypes were found in the INSTINCT cohort, which clustered on the same portals of entry, with two sub-phenotypes in each. When the simplified probabilistic model was applied, the sub-phenotypes showed significant associations with 30-day mortality in both validation cohorts. In INSTINCT, the aHRs were 1·93 (A2 vs A1), 3·40 (B2 vs B1), and 3·04 (C2 vs C1). In FEN-AUREUS, the aHRs were 2·02 (A2 vs A1), 2·11 (B2 vs B1), and 2·44 (C2 vs C1). Interpretation: Patients with SAB can be classified into phenotypes and sub-phenotypes, each exhibiting considerable variations in mortality rates. To facilitate clinical application, a validated open-access algorithm and calculator for phenotype and sub-phenotype assignment have been developed, enabling their use at the time of SAB confirmation. This tool aims to support timely and personalised patient care. Funding: Instituto de Salud Carlos III, Spanish Ministry of Science, Innovation and Universities (PI21/01801).
%K Bacteraemia
%K Clinical profiles
%K Mortality
%K Phenotypes
%K Staphylococcus aureus
%~ Universidad de Cádiz