UP STUDY - Decipher Persistent Critical Illness Through in Deep Clinical Phenotyping.

Persistent Critical Illness (PCI) is a condition that affects some patients who remain in the Intensive Care Unit (ICU) for a long time, usually more than 10-14 days. It is estimated to occur in 5-20% of critically ill patients. A recent Portuguese study found that more than 14% of ICU patients stayed longer than 14 days. PCI is often associated with ongoing need for life support, such as mechanical ventilation or medications to maintain blood pressure. However, patients may also experience severe muscle weakness, repeated infections, or other complications, which makes this group very diverse. One of the main risk factors for prolonged ICU stay is sepsis, a severe infection that affects the whole body. Other factors-such as prior health conditions, use of corticosteroids, sedation practices, early versus late mobilization, fluid and antibiotic management, and delirium treatment-may also influence the development and course of PCI. This study aims to identify different clinical patterns ("clusters") among critically ill patients who remain in the ICU for more than 10 days. Patients will be followed until hospital discharge, and up to one year if data are available. Understanding these different patterns will help develop more personalized and effective care strategies for each patient profile. The study is a multicenter retrospective cohort including adult patients (≥18 years) admitted to participating ICUs for more than 5 days between 2021 and 2023. Data collected will include demographic, clinical, and laboratory information, details of organ support (such as mechanical ventilation or vasopressors), medications, nutrition, and rehabilitation practices. Statistical and machine learning methods will be used to identify groups of patients with similar clinical trajectories and to assess how these groups are related to outcomes such as survival, recovery of organ function, or long-term disability. Expected results are the identification of distinct clinical clust