摘要目的:建立钠-葡萄糖共转运蛋白2抑制剂（SGLT2i）不良反应/事件主动监测模型，供医疗机构应用和推广。方法:研究对象来自广东药科大学附属第一医院（我院）出院时间为2021年3月1日至2022年10月1日并使用SGLT2i的2型糖尿病患者。患者被分为2部分，分别用于预试验阶段和临床应用验证阶段。检索国内外数据库及药品说明书中SGLT2i相关不良反应/事件，初步制订触发条目。经征求我院专家意见，参考相关医嘱、病程记录及实验室指标参考值等，对触发条目进行修改；采用德尔菲法进行问卷调查，根据专家意见对其进行整理、分析讨论及修改，形成初步触发条目。应用中国医院药物警戒系统建立监测模型，通过预试验阶段对触发条目进行完善，通过临床应用阶段对其进行验证。结果:预试验阶段和临床应用验证阶段分别纳入218和858例患者。根据文献和药品说明书初步形成44个触发条目。采用德尔菲法征求专家意见，共回收调查问卷16份，根据专家意见进行修改，并经预试验阶段进一步完善，最终确定24项触发条目，包括实验室指标类（A）8项、解救剂类（B）4项、临床症状类（C）11项和干预措施类（D）1项。预试验阶段和临床应用验证阶段模型监测阳性病例数分别为56和189例，人工审查实际阳性病例数为12和57例。预试验阶段和临床应用阶段触发器的阳性预测值分别为25.0%（18/72）和30.9%（77/249）、不良反应/事件检出率分别为5.5%（12/218）和6.6%（57/858）、灵敏度分别为92.3%（12/13）和100%（57/57）、特异度分别为78.5%（161/205）和83.5%（669/801）。预试验阶段12例阳性病例和临床应用验证阶段57例阳性病例中关联性评价均为很可能和可能（分别为4、8例和16、41例）；不良反应/事件严重程度主要为2级（分别为11例和55例）。SGLT2i不良反应/事件主要为低血糖、尿路感染、皮疹等；胰腺炎、体重下降等在药品说明书中无相关表述，但经关联性评价均为很可能。结论:经预试验和内部临床验证，本研究建立的SGLT2i药品不良反应/事件主动监测模型灵敏度和特异度均较高，可供医疗机构实际应用。

abstractsObjective:To establish an active monitoring model for adverse reactions/events of sodium-glucose cotransporter 2 inhibitor (SGLT2i) for application and promotion in medical institutions.Methods:The subjects were type 2 diabetes patients who were discharged from the First Affiliated Hospital of Guangdong Pharmaceutical University (our hospital) from March 1, 2021 to October 1, 2022 and treated with SGLT2i. The patients were divided into 2 parts and assigned to the pre-trial phase and clinical application validation phase, respectively. SGLT2i-related adverse reactions/events from domestic and foreign databases and drug labels were retrieved, and triggering items were developed preliminarily. After soliciting opinions from experts in our hospital, referring to relevant medical orders, disease course records, and laboratory indicator reference values, the triggering items were modified, and a questionnaire survey was conducted using the Delphi method. According to expert opinions, the items were sorted, analyzed, discussed, and modified to form preliminary triggering items. A monitoring model was established based on the Chinese hospital drug surveillance system, the triggering items were improved during the pre-trial phase, and validated during the clinical application phase.Results:A total of 218 and 858 patients were obtained in the pre-trial phase and clinical application validation phase, respectively. Based on literature and drug labels, 44 triggering items were preliminarily formed. A total of 16 survey questionnaires from experts were collected. After being modified based on expert opinions, and further improved during the pre-trial phase, 24 triggering items were determined finally, including 8 laboratory indicators (A), 4 rescue agents (B), 11 clinical symptoms (C), and 1 intervention measure (D). The number of positive cases monitored by the model during the pre-trial phase and clinical application validation phase was 56 and 189, respectively. The actual number of positive cases under manual review was 12 and 57, respectively. The positive predictive value (PPV) of the triggering items in the pre-trial phase and clinical application phase were 25.0% (18/72) and 30.9% (77/249), respectively, with adverse reaction/event detection rates of 5.5% (12/218) and 6.6% (57/858), sensitivity of 92.3% (12/13) and 100% (57/57), and specificity of 78.5% (161/205) and 83.5% (669/801). Among the 12 positive cases in the pre-trial phase and 57 positive cases in the clinical application validation phase, the association evaluation was probable and possible (4, 8 cases and 16, 41 cases, respectively). The severity of adverse reactions/events was mainly grade 2 (11 cases and 55 cases, respectively). The main adverse reactions/events of SGLT2i were hypoglycemia, urinary tract infections, rashes, etc. Pancreatitis, weight loss, etc., which were not stated in the drug labels, were evaluated as probable.Conclusion:Through pre-trial and internal clinical validation phase, the adverse reaction/event active monitoring model established for SGLT2i in this study has high sensitivity and specificity, and can be applied practically in medical institutions.