目的 构建科学、系统、具有临床适用性的重症患者喂养不耐受预测因子指标体系，为早期识别高风险患者及开发预测模型提供依据。 方法 通过系统评价整合现有预测因子证据，结合两轮德尔菲专家函询筛选指标。纳入27篇文献进行Meta分析，邀请17名重症医学、护理、营养等领域专家，采用Likert 5级评分法评估指标重要性，依据均数、变异系数和满分率进行筛选。 结果 构建了包含6个一级指标、29个二级指标的预测因子体系，涵盖个体特征、疾病严重程度、生化指标、治疗因素、营养支持策略和胃肠功能因素。专家权威系数分别为0.920和0.923，肯德尔和谐系数具有统计学意义（P<0.05）。关键指标包括急性生理与慢性健康评价Ⅱ评分、序贯器官衰竭评估评分、机械通气、腹内压、胃窦运动指数等，均获得高度共识。 结论 该研究系统整合了循证证据与专家共识，构建了具有较高临床适用性的喂养不耐受预测因子体系，为重症患者营养支持的精准化管理提供了科学基础。未来应进一步开展模型开发与临床验证，推动其向实用工具转化。

Objective To construct a scientific，systematic，and clinically applicable predictive indicator system for feeding intolerance in critically ill patients，providing a basis for early identification of high-risk patients and the development of predictive models. Methods Evidence on existing predictive factors was integrated through a systematic review，and indicators were screened via two rounds of Delphi expert consultation. A meta-analysis of 27 studies was conducted，and 17 experts in critical care medicine，nursing，and nutrition were invited to rate the importance of indicators using the Likert 5-point scale. Indicators were selected based on mean score，coefficient of variation，and full-score rate. Results A predictive indicator system comprising 6 first-level indicators and 29 second-level indicators was established，covering individual characteristics，disease severity，biochemical indicators，treatment factors，nutritional support strategies，and gastrointestinal function factors. The expert authority coefficients were 0.920 and 0.923，and the coordination coefficients were statistically significant（P<0.05）. Key indicators，including APACHE Ⅱ score，SOFA score，mechanical ventilation，intra-abdominal pressure，and antral motility index，achieved high consensus. Conclusion This study systematically integrated evidence-based findings and expert consensus to develop a clinically applicable predictive indicator system for feeding intolerance in critically ill patients，providing a scientific foundation for precise nutritional management. Future research should focus on model development and clinical validation to translate the system into practical tools.