Temperature Trajectory Subphenotypes in Oncology Patients with Neutropenia and Suspected Infection

This study aimed to determine if temperature trajectory modeling could identify distinct subgroups of oncology patients with neutropenia and suspected infection. The study used a validated model to classify patients' temperature patterns over the first 72 hours into four subphenotypes. The study included a primary cohort of 1,145 patients and a validation cohort of 6,564 patients. The four temperature subphenotypes identified were "hyperthermic slow resolvers," "hypothermic," "hyperthermic fast resolvers," and "normothermic." The subphenotypes "hyperthermic slow resolvers" and "hypothermic" had higher mortality rates compared to "hyperthermic fast resolvers" and "normothermic." Bloodstream infections were more common in the "hyperthermic slow resolvers" and "hyperthermic fast resolvers" groups compared to the "hypothermic" and "normothermic" groups. After adjusting for confounding factors, the study found that "hyperthermic slow resolvers" had increased odds of mortality and bloodstream infection in both the primary and validation cohorts. In conclusion, temperature trajectory subphenotypes were independently associated with important outcomes in hospitalized oncology patients with neutropenia in this study, which was conducted across two independent cohorts.