目的: 基于DRG-DIP映射关系评估上海按病种付费2.0版分组效能,探索分组优化路径,为深化“双D融合”提供实证依据。方法: 利用上海市2021—2023年住院结算病案数据,构建DRG与DIP映射关系,采用变异系数（CV）与方差减少系数（RIV）分别评价组内同质性与可细分潜力,并建立四象限分析框架进行分组效能分类。选取典型病组,依据操作类型或主诊/主操开展细分验证。结果: 方案整体分组效能较好,73.91%的病组CV≤1,90.12%的病组RIV≤0.4;传统DRG规则组与底层聚类组的同质性不达标率分别为31.11%和22.52%。共识别出31个高异质性可细分组,其异质性主要缘于手术复杂度、解剖部位、肿瘤良恶性及高值耗材混杂等因素。案例分析显示,外科组应以操作复杂度分层为优化核心,底层聚类组需引入诊断维度（如肿瘤良恶性）,内科组在编码层面的细分空间有限。结论: DRG-DIP映射为分组效能评估提供了新视角。上海2.0版分组方案整体合理性较好,分类施策是其进一步优化的关键路径,该方案或可为全国“双D融合”探索提供参考。

Objective: The paper attempts to evaluate the grouping effectiveness of Shanghai’s Case-Based Payment Version 2.0 and to explore optimization pathways based on the DRG-DIP mapping, providing empirical evidence for advancing the DRG-DIP integration. Methods: Using inpatient discharge settlement data in Shanghai from 2021 to 2023, a mapping relationship between DRG and DIP was constructed. The coefficient of variation (CV) and reduction in variance (RIV) were applied to assess within-group homogeneity and subdivision potential. A four-quadrant analytical framework was established to classify grouping effectiveness. Typical disease groups were selected, and detailed validation was conducted based on procedure types or principal diagnosis/principal procedure. Results: The overall grouping performance was satisfactory: 73.91% of groups met the homogeneity criterion (CV ≤ 1), and 90.12% showed limited withingroup structural variation (RIV ≤ 0.4). The proportion of groups failing to meet homogeneity standards was respectively 31.11% under traditional DRG rules and 22.52% under the underlying clustering approach. A total of 31 high-heterogeneity groups with subdivision potential were identified, with heterogeneity mainly driven by surgical complexity, anatomical site, benign or malignant nature of the tumor, and the use of high-cost consumables. Case analyses indicated that optimization of surgical groups should focus on stratification by procedural complexity, while clustering-based groups require the incorporation of diagnostic dimensions (such as benign or malignant nature of the tumor). In contrast, the scope for further refinement in medical (non-surgical) groups remains limited at the coding level. Conclusion: DRG-DIP mapping provides a novel perspective for evaluating grouping effectiveness. The Shanghai Version 2.0 is generally reasonable, and classification-based optimization is key to its further improvement. This approach may serve as a reference for the nationwide exploration of DRG-DIP integration.