Identification and Characterization of Sepsis Phenotypes in an Indian Cohort.

Abstract

Background: Sepsis has a heterogenous clinical presentation with distinct regional epidemiological profiles. The study aims to identify clinical phenotypes from an Indian sepsis cohort.

Methods: The cohort included all adult patients admitted through the emergency department with a presumed diagnosis of sepsis following Surviving Sepsis Guidelines criteria. The unsupervised k-means clustering method was used, and the identified Indian clusters were compared using distance matching to phenotypic clusters in Seymour et al., which had previously identified four reproducible sepsis phenotypes in large Western ICU cohorts. The association of focus of infection and outcomes among the clusters were assessed.

Results: Of the 1009 patients, four clusters were identified. Age (p = 0.024) and gender (p = 0.04) significantly differed among the clusters. The Indian clusters that exhibited close association to the alpha, beta, gamma, and delta phenotypes reported in Seymour et al. were termed i-Alpha (254, 25%), i-Beta (141, 14%), i-Gamma (389, 39%), and i-Delta (225, 22.3%) respectively. The clusters demonstrated significant variations in clinical profiles and outcomes. Septic shock and in-hospital mortality were highest among i-Alpha at 19% (48/254) and i-Beta at 34% (48/141) (p < 0.001), respectively. Prevalence of pneumonia (p = 0.018), SSI (p = 0.011), and bacteremia (p = 0.004) significantly differed among the clusters. Pseudomonas (p = 0.008) and Staphylococcus aureus (p = 0.02) were also observed to be significantly different. i-Gamma was characterized by a lower mortality rate and a high prevalence of soft tissue infections.

Conclusion: Understanding these unique phenotypes can guide personalized treatment and improve sepsis outcomes in resource-limited settings, highlighting the importance of localized research in sepsis management.