# Unmasking Heterogeneity: Evaluating Distinct Sepsis Clinical Phenotypes and Their Association With Mortality in Critical Adult Patients

## TL;DR

This study identifies distinct sepsis patient groups based on clinical traits and finds that some groups face higher mortality risks.

## Contribution

The study introduces and evaluates three novel phenotypic classifications for sepsis patients to predict mortality.

## Key findings

- Elderly patients with hypothermia had a 17-fold higher mortality risk compared to younger or febrile patients.
- Patients with multi-organ failure had a 6-fold higher mortality risk than others.
- Combined phenotypic models predicted mortality with 85.6% accuracy, outperforming APACHE II and SOFA.

## Abstract

Objective: This study aimed to evaluate three phenotypic classifications based on their ability to predict mortality in critically ill adult patients with sepsis.

Methods: This single-center cohort study involved 106 patients diagnosed with sepsis upon admission to the intensive care unit (ICU). The patient population was categorized according to three distinct clinical phenotypic models: the first based on age and thermal profiles, the second stratifying patients into four specific groups (multiple organ failure, respiratory dysfunction, neurological dysfunction, and miscellaneous), and the third assessing arterial blood pressure trajectories within the initial 10 hours of ICU stay. Multiple logistic regression models were fitted to evaluate the utility of these classifications in predicting in-hospital mortality. The performance of a model incorporating the three phenotypic classifications was compared with the acute physiology and chronic health evaluation II (APACHE II) and sequential organ failure assessment (SOFA) scores.

Results: It was observed that elderly patients presenting with hypothermia showed a significantly higher mortality risk compared to younger, normothermic, or febrile subjects (OR 17.32 [95% CI 1.95-153.14], p = 0.010). Furthermore, patients with multi-organ failure presented a significantly higher risk of death (OR 5.87 [95% CI 1.17-29.94], p = 0.031). Finally, persistent hypotensive individuals were not found to have a significantly elevated risk of death (p = 0.300). The final predictive model’s area under the ROC curve was 0.856, which was not inferior to that of APACHE II (0.776) or SOFA (0.764) in the sample studied.

Conclusions: When used together, the phenotypically analyzed classifications demonstrated good accuracy in predicting mortality among critically ill patients with sepsis.