Abstract: Background　Growing prevalence of diabetes contributes to increasing attentions given to inpatient glycemic management. However，the lack of recognized definition and measurement of performance indicators，as well as the standardized blood glucose benchmark report has affected the continuous quality improvement of glycemic management in hospitals. Objective　To develop a blood glucose benchmark report based on an information glucose monitoring system （IGMS），providing methodological guidance for developing such reports，and a blood glucose benchmark for medical workers for reference. Methods　Patients with diabetes or hyperglycemia （no history of diabetes but with peripheral blood glucose levels exceeding 11.1 at any random time） were selected during October 2019 to March 2020 from non-intensive care units （10 medical and 7 surgical units） equipped with IGMS （for automatically uploading blood glucose data and reviewing blood glucose data on demand），Suining Central Hospital. The blood glucose data on the first day of hospitalization were excluded. Three models of population，patient and patient-day were used to report the incidence of ideal blood glucose，hyperglycemia and hypoglycemia. The ratio of days，frequency and management timeliness of patients were additionally added for hypoglycemia report. Our internal glucose data were analyzed by quartiles and compared with the blood glucose data of medical institutions in the United States，Australia，Chongqing and Guangdong Province. Results　The average blood glucose，ideal blood glucose ratio，and average hyperglycemia differed significantly across the three models（P<0.001）. The ratios of hyperglycemia （>10.0 mmol/L，≥15.0 mmol/L，≥16.7 mmol/L），and hypoglycemia （≤3.9 mmol/L，<3.0 mmol/L，<2.8 mmol/L，<2.2 mmol/L） at any random time of the three models were statistically different （P<0.001）. Of the patients in 17 wards of our hospital，312 cases （3.9%） had hypoglycemia （≤3.9 mmol/L） occurring more than 3 times at any random time，202（2.5%） had hypoglycemia occurring more than 3 days，446（5.6%） remeasured blood glucose ≤15 minutes，and 2 187（27.5%） remeasured blood glucose >30 minutes. The patient-day percentages of patients in lower and upper quartiles of hyperglycemia （≥15.0 mmol/L） at any random time were 38.6% and 21.5%，respectively. The patient-day percentages of patients in lower and upper quartiles of hypoglycemia（≤3.9 mmol/L） at any random time were 5.5% and 2.1%，respectively. The patient-day percentages of patients in lower and upper quartiles of ideal blood glucose were 38.0% and 58.1%，respectively. Statically significant differences（P<0.05） were found in average hyperglycemia，hyperglycemia at any random time（>10.0 mmol/L，≥15.0 mmol/L，≥16.7 mmol/L），and hypoglycemia at any random time（≤3.9 mmol/L，<3.0 mmol/L，<2.8 mmol/L，<2.2 mmol/L） across diabetic or hyperglycemic patients in our 17 wards，and in medical institutions of the United States，Australia and Chongqing. The average blood glucose，ideal blood glucose，hyperglycemia （≥16.7 mmol/L） at any random time，and hypoglycemia （≤3.9 mmol/L） at any random time in diabetic or hyperglycemic patients in the Endocrine Ward of our hospital were significantly different from those of diabetic or hyperglycemic patients in an Endocrine Ward in Guangdong Province（P<0.05）. Conclusion　IGMS contributes to the creation of comprehensive and standardized glucose report in medical institutions. The selection of model varies by the purpose of creation：the population model is more suitable for drug management and risk management for individuals，the patient model is more suitable for individualized nursing evaluation，and the patient-day model is more suitable for deciding the quality direction and goal to be improved for a unit.

Key words: Diabetes mellitus, Blood glucose monitoring, Ideal blood glucose, Hypoglycemia, Hyperglycemia, Glucose benchmark report

摘要： 背景　随着糖尿病患病率的增高，住院患者血糖管理已成为医院关注的焦点，但目前尚缺乏公认的绩效指标定义、指标测算和标准化血糖基准报告，影响了医院血糖管理持续质量改进。目的　基于信息化血糖监测系统（IGMS）构建血糖基准报告，一为同行建立标准化血糖报告提供方法学指导，二为同行提供血糖基准参考。方法　纳入2019年10月至2020年3月遂宁市中心医院安装了IGMS（该系统能自动上传血糖数据和按需求查阅血糖数据）的非重症监护病区（10个内科和7个外科）的糖尿病患者或高血糖患者（无糖尿病病史但随机末梢血糖超过11.1 mmol/L），排除住院第1天的血糖数据。采用群体（population）、患者（patient）和患者日（patient-day）三种模型报告理想血糖、高血糖和低血糖发生率，低血糖报告增加患者低血糖发生天数比、发生频次比和管理及时性比，采用四分位数分析不同病区的血糖数据，并与美国、澳大利亚、重庆和广东医疗机构的血糖数据进行比较。结果　三种模型平均血糖、理想血糖、平均高血糖和任意高血糖（>10.0 mmol/L、≥15.0 mmol/L、≥16.7 mmol/L）、任意低血糖（≤3.9 mmol/L、<3.0 mmol/L、<2.8 mmol/L、<2.2 mmol/L）比较，差异均有统计学意义（P<0.001）。17个病区任意低血糖≤3.9 mmol/L发生>3次者312例（3.9%），发生>3 d者202例（2.5%），复测血糖时间≤15 min者446例（5.6%），>30 min者2 187例（27.5%）。患者日任意高血糖≥15.0 mmol/L、任意低血糖≤3.9 mmol/L、理想血糖下四分位数分别为38.6%、5.5%和38.0%，上四分位数分别为21.5%、2.1%和58.1%。17个病区与美国、澳大利亚、重庆医疗机构的平均高血糖和任意高血糖（>10.0 mmol/L、≥15.0 mmol/L、≥16.7 mmol/L）、任意低血糖（≤3.9 mmol/L、<3.0 mmol/L、<2.8 mmol/L、<2.2 mmol/L）比较，差异均有统计学意义（P<0.05）。本研究内分泌病区和广东省某内分泌病区血糖、理想血糖、任意高血糖≥16.7 mmol/L、任意低血糖≤3.9 mmol/L比较，差异均有统计学意义（P<0.05）。结论　IGMS允许医疗机构建立全方位、标准化血糖报告。因目的不同，其选择模型也不同。群体模型更适用药物管理和风险管理人员，患者模型更适用个体化护理评估，患者日模型适用于某单元或单位进行质量改进方向和目标的制定。

关键词: 糖尿病, 血糖监测, 理想血糖, 低血糖, 高血糖, 血糖基准报告