Prognosis Prediction in Adult Intensive Care—A Critical Analysis

Abstract

Background: Prognosis prediction in adult intensive care units (ICUs) guides clinical decision-making, realistic goal-setting with families, and resource allocation, but is challenged by the marked heterogeneity of critically ill patients in multi-speciality wards. Methods: This narrative review critically analyses commonly used prognostic scoring systems by clinical application timeline. Results: Admission prognostic models, including the Acute Physiology and Chronic Health Evaluation (APACHE II/III/IV), Simplified Acute Physiology Score (SAPS II/SAPS 3) and Mortality Probability Model (MPM), estimate in-hospital mortality from acute physiological derangement and comorbidity; the Clinical Frailty Scale (CFS) adds prognostic value. During the ICU stay, organ dysfunction is tracked with the Sequential Organ Failure Assessment (SOFA/SOFA-2), designed for repeated reassessment, and the Multiple Organ Dysfunction Score (MODS), while the Therapeutic Intervention Scoring System (TISS-28) quantifies nursing workload. At discharge, the Barthel Index, Glasgow Outcome Scale (GOS) and Cerebral Performance Category (CPC) measure functional independence and neurological outcome. Admission-based mortality models use an initial assessment window and may require local recalibration, and discrimination alone does not establish clinical utility. Conclusions: Future directions include trajectory-based models, phenotyping, biomarkers, time-limited trials and machine learning. Prognostic estimates should inform multidisciplinary evaluation and shared decision-making, rather than independently determine treatment intensity or withdrawal.