Multicenter Validation of Clinical Sepsis Phenotypes
Authors' conclusion
This study suggests that the 4 clinical phenotypes of the SENECA data are not generalizable across 3 independent cohorts.
Study context
- Population
- Adults admitted to university hospital EDs with body fluid cultures, antibiotics and SOFA score ≥2, all within 6 hours of admissionpatient encounters
- Participants
- 30,865 / 15,575 / 1,806
- Exposure
- Site-derived clinical phenotypes vs Sepsis Endotyping in Emergency Care (SENECA)-derived phenotypes, and between sites
- Stockholmn = 30,865
- Oxfordn = 15,575
- Oslon = 1,806
- Outcome
- Consistency of 4 k-means consensus-clustering clinical phenotypes derived at each site with SENECA-derived phenotypes and with one another
Methodological detailsmulticentre, validation study, consensus clustering with k-means, Data analysis November 2, 2023, to September 1, 2025
- Multicentre
- Validation study
- Consensus clustering with k-means
- Data analysis November 2, 2023, to September 1, 2025
Main finding
As reported in the abstract
Agreement with SENECA clinical phenotypes
| Group | Cohen κ | Adjusted Rand Index |
|---|---|---|
| Stockholm | 0.32 | 0.21 |
| Oxford | 0.40 | 0.26 |
| Oslo | 0.37 | 0.27 |
Cohen κ
Stockholm0.32
Oxford0.40
Oslo0.37
Adjusted Rand Index
Stockholm0.21
Oxford0.26
Oslo0.27
Observational data: associations, not effects.
Further findings
Consistency between phenotypes derived in Stockholm, Oxford and Oslo
- Little consistency between the sites' phenotypes
Original publication
Multicenter Validation of Clinical Sepsis Phenotypes. JAMA Netw Open. Published 1 June 2026. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →