Background Gastrointestinal ultrasonography (GI-US) has significant potential for predicting enteral nutrition intolerance (ENFI) in critically ill adults. However, current evidence is fragmented, with substantial variability in predictors, diagnostic thresholds, acquisition protocols, and a lack of cross-indicator comparisons. These inconsistencies hinder evidence synthesis and the standardization of clinical practice. Objective This study aimed to systematically synthesize and evaluate the diagnostic performance of GI-US predictors for ENFI, identifying core predictors that balance accuracy with clinical feasibility. Methods Comprehensive searches were conducted in PubMed, Web of Science, Embase, CINAHL, the Cochrane Library, CNKI, Wanfang, and SinoMed through September 12, 2024. Two independent reviewers screened studies, extracted data, and assessed quality using the QUADAS-2 tool. Diagnostic accuracy was analyzed using a bivariate random-effects model to generate summary estimates of sensitivity, specificity, likelihood ratios, diagnostic odds ratios, and SROC-AUCs, all with 95% confidence and prediction intervals. Threshold effects, heterogeneity, sensitivity analyses, publication bias (via Deeks’ test), and clinical utility (via Fagan nomograms) were comprehensively evaluated. Results The analysis included 16 studies with 1,261 participants, evaluating ten single predictors, one composite score, and two multivariable models. Seven studies ( n = 502) on gastric antrum cross-sectional area (GCSA) were suitable for meta-analysis, demonstrating good diagnostic accuracy (SROC-AUC 0.86, 95% CI 0.76–0.89; sensitivity 0.82, 95% CI 0.76–0.88; specificity 0.77, 95% CI 0.72–0.81), with no significant heterogeneity or publication bias. At a pretest probability of 42.91%, a positive GCSA result increased the posttest probability of ENFI to 72.51%, while a negative result decreased it to 14.74%. Other single predictors varied in accuracy, whereas composite scores and multivariable models achieved higher AUCs (up to 0.95) but were supported by limited evidence. Conclusion Current GI-US predictors can be categorized into gastric, intestinal, and superior mesenteric artery parameters. GCSA is supported by strong evidence for quantitative synthesis, demonstrating reliable diagnostic accuracy and bedside feasibility. Other predictors and emerging models show promise but require validation. Large, standardized multicenter studies are needed to validate multidomain predictors, harmonize measurement protocols, and develop integrated frameworks for effective ENFI management. Systematic review registration PROSPERO , Identifier CRD420251019447.
Jia et al. (2026) studied this question.