Development and Validation of a Multidimensional Ultrasound-based Gastrointestinal Congestion Score (GICS) for Predicting Persistent Acute Gastrointestinal Injury in Critically Ill Patients

Technical Summary: Development and Validation of the Gastrointestinal Congestion Score (GICS)

Problem Statement
Acute gastrointestinal injury (AGI) is a prevalent and severe complication in critically ill patients, associated with prolonged intensive care unit (ICU) stays, increased costs, and higher mortality. Current diagnostic guidelines rely heavily on subjective clinical signs (e.g., abdominal distension, feeding intolerance), which often manifest only after significant mucosal damage has occurred. While venous congestion is recognized as a primary driver of organ dysfunction, traditional assessment methods face limitations. Specifically, the widely used Venous Excess Ultrasound (VExUS) grading system relies on inferior vena cava (IVC) dilation as a mandatory component. However, the gastrointestinal tract possesses a highly compliant venous reservoir that may harbor congestion even when systemic venous pressures remain normal. Factors such as elevated vascular tone, intra-abdominal hypertension, or specific hemodynamic states can restrict IVC distensibility while simultaneously exacerbating splanchnic venous backpressure. This creates an "occult congestion" phenotype—significant visceral stasis with a non-dilated IVC—which traditional IVC-dependent protocols fail to detect, leading to delayed identification of patients at risk for persistent AGI.

Methodology
This study conducted a secondary analysis of pooled data from three prospective observational cohorts involving 88 critically ill adults admitted to an Emergency ICU. Patients were assessed within 24 hours of admission using point-of-care ultrasound (POCUS) to evaluate three specific hemodynamic parameters:

1. Inferior Vena Cava (IVC) Diameter: Categorized as normal (<2.0 cm) or dilated (≥2.0 cm).
2. Hepatic Venous Doppler Pattern (HVDP): Graded from 0 (normal, S>D) to 2 (severe, S-wave reversal).
3. Portal Venous Doppler Pattern (PVDP): Graded based on the Pulsatility Index (PI), ranging from 0 (PI <30%) to 2 (PI ≥50% or interrupted flow).

The primary outcome was persistent AGI, defined as the failure to achieve enteral nutrition goals for ≥3 consecutive days due to gastrointestinal symptoms. The study utilized Least Absolute Shrinkage and Selection Operator (LASSO) regression for variable selection and multivariable logistic regression to construct predictive models. Three models were compared: a baseline Clinical Model, an Unintegrated Ultrasound-Clinical Model (incorporating raw ultrasound metrics), and a novel GICS-Clinical Model (incorporating the derived Gastrointestinal Congestion Score). Model performance was evaluated using Receiver Operating Characteristic (ROC) analysis, internal bootstrap validation, calibration plots, and Decision Curve Analysis (DCA).

Key Contributions
The paper introduces the Gastrointestinal Congestion Score (GICS), a multidimensional, organ-specific scoring system designed to quantify splanchnic venous burden more accurately than systemic indices.

• Score Construction: The GICS (range 0–7) assigns weighted points based on the severity of IVC dilation, HVDP, and PVDP abnormalities. The weighting was derived from regression coefficients but simplified into an ordinal integer scale (IVC: 0/1; HVDP: 0/2/4; PVDP: 0/1/2) to enhance bedside applicability while maintaining discriminative power.
• Identification of "Occult Congestion": The study highlights a specific phenotype where patients exhibit normal IVC diameters but abnormal splanchnic Doppler patterns. This subgroup (22.7% of the cohort) demonstrated a 92.5% incidence of persistent AGI, underscoring the failure of IVC-only protocols to capture localized gastrointestinal stasis.
• Model Integration: The authors developed a streamlined nomogram (GICS-Clinical Model) that integrates the GICS with key clinical variables (Day 1 AGI grade, SOFA score, Diabetes Mellitus status, and Mechanical Ventilation requirement).

Results

• Prevalence: Persistent AGI occurred in 51 patients (58.0%). The "occult congestion" phenotype was identified in 20 patients (22.7%).
• Discriminative Ability: The GICS alone achieved an Area Under the Curve (AUC) of 0.812, significantly outperforming the simplified VExUS score (AUC 0.654), isolated HVDP (0.649), PVDP (0.675), and IVC diameter (0.644).
• Predictive Models: The GICS-Clinical Model demonstrated exceptional discriminative capacity with an AUC of 0.953 (95% CI: 0.915–0.990). This performance was statistically comparable to the more complex Unintegrated Ultrasound-Clinical Model (AUC 0.954) and significantly superior to the baseline Clinical Model (AUC 0.866).
• Risk Factors: Multivariable analysis confirmed that each 1-point increment in GICS was associated with a nearly 7-fold increase in the odds of persistent AGI (OR = 6.82). Mechanical ventilation and higher initial AGI grades were also significant independent predictors, while Diabetes Mellitus appeared as a protective factor.
• Validation: Internal validation via 1,000-iteration bootstrap resampling confirmed the robustness of the estimates, and calibration plots indicated high fidelity between predicted probabilities and observed outcomes.

Significance and Claims
The authors claim that the GICS provides an objective, organ-specific assessment of gastrointestinal venous congestion that addresses the limitations of systemic VExUS grading. By capturing "occult" splanchnic stasis frequently missed by traditional IVC-dependent protocols, the GICS-Clinical Model enables the early, non-invasive identification of patients at high risk for persistent AGI. The study posits that this multidimensional approach facilitates timely and personalized de-resuscitation strategies and optimization of mechanical ventilation parameters. The authors emphasize that while the model shows high predictive accuracy, it is currently an exploratory risk-stratification tool requiring external validation in diverse, multicenter cohorts before widespread clinical adoption. They explicitly state that the high performance is partly driven by the inclusion of the initial AGI grade, but the GICS retains significant incremental prognostic value beyond baseline clinical assessment.