FOUR-Based Modification of ICU Severity Scores

Overview

The goal of this observational study is to evaluate whether the Full Outline of UnResponsiveness (FOUR) score can replace the Glasgow Coma Scale (GCS) as the neurologic component of established intensive care unit (ICU) severity scoring systems and to derive and validate modified APACHE, SAPS, MODS, LODS, and SOFA models incorporating the FOUR score in critically ill adult patients. The main questions it aims to answer are: Does replacing the GCS with the FOUR score provide comparable or improved discrimination for in-hospital mortality across established ICU severity scoring systems? Can FOUR-based modifications of APACHE, SAPS, MODS, LODS, and SOFA be successfully derived and validated with adequate discrimination, calibration, and overall predictive performance? Do the modified FOUR-based models demonstrate predictive performance comparable to or different from the corresponding original GCS-based models? Researchers will compare the performance of the modified FOUR-based models with their original GCS-based counterparts to determine whether substitution of the FOUR score provides valid and clinically useful risk prediction in critically ill adults. Participants will not receive an experimental intervention. Data routinely collected during ICU care will be used to: Calculate the original GCS-based APACHE, SAPS, MODS, LODS, and SOFA scores. Replace the GCS component with the FOUR score and derive the corresponding modified models. Evaluate and validate the predictive performance of the original and modified models using measures of discrimination, calibration, and overall predictive accuracy.

Intensive care units (ICUs) are highly complex, resource and technology intensive settings. During the 1970s and 1980s, increasing intensive care costs and poor outcomes among patients with multiple organ failure created a need for tools to measure illness severity and patient outcomes. Intensive care unit mortality varies considerably worldwide, with rates of 16.2% in large international study and 19.1% in a European cohort study, while higher mortality was reported in developing countries, Ethiopia (39.06%) and Iraq (29.8%). The Glasgow Coma Scale (GCS) assesses impaired consciousness using eye opening, verbal, and motor responses and is widely used for assessment, triage, monitoring, and communication among healthcare professionals. It has several limitations, verbal component cannot be assessed reliably in patients who are intubated, pharmacologically sedated, or unable to speak, limiting the use of the total score for prognosis. Also has limitation in evaluation patients with concomitant head and spinal cord injury. Drug and alcohol intoxication can also affect GCS assessment and should be considered when interpreting scores. In addition, Aphasia, paralysis, neuromuscular weakness, and spinal injuries may impair verbal or motor responses independently of consciousness, while painful stimuli can complicate scoring in polytrauma patients. Its reliability depends on assessor training, the patient's consciousness level, and the type of stimulus used. The Full Outline of UnResponsiveness (FOUR) score consisting of four components: eye response, motor response, brainstem reflexes, and respiration, with each component assigned a maximum score of 4. It showed excellent interrater reliability (weighted kappa = 0.82), and provide greater neurological detail and identification of severe impairment than GCS. In non-traumatic patients, FOUR score showed better mortality discrimination than GCS (AUROC 0.799 vs. 0.731), and offers advantages in intubated patients because it evaluates brainstem reflexes and respiratory patterns. FOUR also showed better mortality prediction than GCS in traumatic brain injury, neurosurgical, and acute brain-injury patients, with AUROC values of 0.93 vs. 0.89, 0.768 vs. 0.699, and 0.835 vs. 0.772, respectively. The Acute Physiology and Chronic Health Evaluation (APACHE) assess risk of in-hospital by using pre-admission health status and 34 physiological variables collected during the first 24 hours of ICU admission. The Acute Physiology and Chronic Health Evaluation II (APACHE II) uses 12 physiological variables, along with age and previous health status, to calculate a score from 0-71, with higher scores indicating greater illness severity and mortality risk. APACHE III incorporated over 100 variables, while APACHE IV added further clinical factors and enabled prediction of hospital mortality and ICU length of stay. The Simplified Acute Physiology Score (SAPS) includes14 physiological variables collected during the first 24 hours of ICU admission to provides a simple measure of the risk of death among critically ill patients. The Simplified Acute Physiology Score II (SAPS II) provide an objective assessment of illness severity and an estimate of hospital mortality among critically ill patients. SAPS II involve 17 variables, including 12 physiological measurements, age, type of ICU admission, and three underlying disease conditions: acquired immunodeficiency syndrome, metastatic cancer, and hematologic malignancy. SAPS 3, incorporated 20 variables and provided region-specific mortality prediction for international use. The Multiple Organ Dysfunction Score (MODS) assesses six organ systems using PaO₂/FiO₂ ratio, creatinine, bilirubin, platelet count, Glasgow Coma Scale, and pressure-adjusted heart rate. Each variable is scored from 0-4, producing a total score of 0-24, with higher values associated with increased ICU mortality. The Sequential Organ Failure Assessment (SOFA) assesses organ dysfunction and illness severity, particularly in sepsis. It evalutes six organ systems: respiratory function using the PaO₂/FiO₂ ratio, coagulation using platelet count, liver function using serum bilirubin, cardiovascular function according to mean arterial pressure and vasopressor requirements, central nervous system function using the Glasgow Coma Scale, and renal function using serum creatinine or urine output. The worst daily values are scored from 0-4, giving a total of 0-24, with higher scores indicating greater organ dysfunction. SOFA is practical for bedside use and is mainly used to quantify organ dysfunction, evaluate treatment effects, and monitor progression, with higher scores are associated with increased mortality in sepsis. The Logistic Organ Dysfunction System (LODS) was derived from 12 countries. It evaluates six organ systems using variables related to neurologic, cardiovascular, renal, pulmonary, hematologic, and hepatic function, producing a score from 0-22. It showed good discrimination, with AUROC values of 0.843 in the development sample and 0.850 in the validation sample. It can also accurately describe the progression of illness when applied on a daily basis. Trauma severity scoring systems help assess injury severity, guide triage, and predict patient outcomes. It can be divided into anatomical, physiological, and combined scores each with advantages and disadvantages. Anatomical scores describe the injuries recorded by clinical examination, imaging, surgery, or autopsy. While Physiologic Scores describe changes due to a trauma and translated by changes in vital signs and consciousness. Combining anatomical and physiological scores may provide better prediction of mortality in severely injured ICU patients. The Trauma and Injury Severity Score (TRISS) combines Revised Trauma Score, Injury Severity Score, and patient age, to improve prediction of trauma outcomes. The Rapid Emergency Medicine Score (REMS) is a physiological score derived from APACHE II and provides rapid risk stratification using age, mean arterial pressure, respiratory rate, pulse rate, oxygen saturation, and Glasgow Coma Scale. The modified REMS (mREMS) substituting systolic blood pressure for mean arterial pressure, reducing the weighting of age, and increasing the weighting of GCS. It showed an AUROC of 0.967 for predicting in-hospital mortality. Neurologic status is central to ICU prognosis and risk adjustment, but the Glasgow Coma Scale (GCS) has practical limitations because its verbal component is difficult or impossible to assess in intubated or mechanically ventilated patients. The FOUR score was developed to address this by using eye, motor, brainstem reflexes, and respiration, and it has shown feasibility and reliable use in emergency and intensive care settings. Prior studies suggest FOUR is often at least comparable to GCS for mortality prediction and may perform better in some ICU cohorts, especially when verbal assessment is compromised. Despite this promise, the available evidence largely evaluates FOUR as a standalone score, not as a direct replacement for GCS inside established ICU severity models. No retrieved studies identified substantive derivation, recalibration, or external validation of FOUR-substituted versions of APACHE, SAPS, MODS, or LODS. The only identified model-level substitution evidence was for SOFA, and it was unfavorable: a FOUR-based modification did not improve mortality prediction and showed lower AUROC than the standard GCS-based SOFA. This leaves a clear methodological gap in whether FOUR can be validly integrated into legacy severity scores rather than only used as an independent neurologic assessment. Accordingly, this study aims to determine whether FOUR can replace GCS in established ICU severity scores and whether such substitution supports the derivation and validation of modified APACHE, SAPS, MODS, LODS, and SOFA models.