深圳1型糖尿病联盟登记数据质量评价研究:基于中国1型糖尿病联盟协作研究平台网

Quality evaluation of registration data of Shenzhen Type 1 Diabetes Alliance:based on the collaborative research platform network of China Type 1 Diabetes Alliance

目的对深圳1型糖尿病联盟(简称"深盟T1D")登记的数据质量进行评价,为持续提高数据录入质量提供评估与改进依据。方法纳入2018年12月至2021年7月间,由深盟T1D登记于中国1型糖尿病联盟协作研究平台网(简称"中国T1D网")的首诊1型糖尿病(T1DM)患者697例(深圳市内501例和市外196例)和复诊T1DM患者120例(深圳市内113例和市外7例)。从数据完成度、准确度和复诊情况三个维度对数据质量进行评价。数据完成度采用整体数据完成度和关键指标完成度进行评价;数据准确度采用血糖异常值概率进行评价;患者复诊情况利用复诊率进行评价。结果深盟T1D首诊T1DM患者特征以体型偏瘦[BMI(19.80±3.52)kg/m2]、青壮年[年龄(34.4±17.1)岁]为主,男女患者约各占一半[男性比例:52.4%(365/697)],确诊类型以经典T1DM[65.22%(150/230)]和成人隐匿性自身免疫性糖尿病(LADA)[26.08%(60/230)]为主,患者空腹血糖(FPG)[(10.93±6.98)mmol/L]和糖化血红蛋白(HbA_(1c))[(10.63±3.01)%]偏高。深盟T1D首诊T1DM患者整体数据完成度仅六成[(62.9±31.5)%]:填报完成度≥80%的首诊患者仅有一半[50.2%(350/697)],深圳市内整体数据完成度≥80%的患者较市外少[44.3%(222/501)vs 65.3%(128/196),P<0.001]。首诊关键指标完成度较好的有病程[76.2%(531/697)]、起病年龄[75.8%(528/697)]和糖尿病家族史[74.9%(522/697)]等,但均无完成度超过八成者,其中糖尿病自我管理行为评估问卷和量表得分两关键指标完全缺失;深圳市内和市外数据对比显示,日常血糖监测频率[46.1%(231/501)vs 64.3%(126/196),P<0.001]、目前胰岛素方案[44.3%(222/501)vs 63.3%(124/196),P<0.001]、起病至今糖尿病酮症酸中毒(DKA)次数[45.7%(229/501)vs 64.8%(127/196),P<0.001]及近1个月症状性低血糖次数[39.3%(197/501)vs 63.8%(125/196),P<0.001]等关键指标以市外填报完成度更高。此外,FPG和餐后2 h血糖(PPG)数据异常值概率较低[5.2%(24/466);3.8%(9/236)];复诊率不高[17.2%(120/697)],其中市内高于市外[22.6%(113/501)vs 3.6%(7/196),P<0.001];首次复诊率为16.2%(113/697),二次复诊率严重不足[1.0%(7/697)]。结论深盟T1D录入T1DM患者数据质量有待进一步提高。通过完善中国T1D网与深盟T1D的信息互联,以及配备专职质控人员和构建系统化的数据质量评估分析和反馈机制,或可促进T1DM数据全面准确高效录入、促进评估方法持续改进以及提高整体数据质量。

Objective To evaluate the data quality of Shenzhen Type 1 Diabetes Alliance(SZT1D),and to provide a basis for evaluation and improvement for the continuous improvement of data quality.Methods From December 2018 to July 2021,697 first-visit type 1 diabetes(T1DM)patients(including 501 in Shenzhen and 196 out-of-Shenzhen)and 120 re-visited T1DM patients(including 113 in Shenzhen and 7 out-of-Shenzhen)who were registered by SZT1D in collaborative research platform network of China Type 1 Diabetes Alliance(hereinafter referred to as China T1D).The data quality was evaluated from three dimensions:data completion,accuracy and revisit.The data completion degree was evaluated by the overall data completion degree and the key indicator completion degree;the data accuracy was evaluated by the probability of abnormal blood glucose value;the patient′s return visit was evaluated by the return visit rate.Results The main characteristics of T1DM in SZT1D were young and middle-aged adults[age:(34.4±17.1)years]with thin body[BMI:(19.80±3.52)kg/m2)],half of male and female patients[proportion of male:52.4%(365/697)];the main types of diagnosis were classical T1DM[65.22%(150/230)]and latent autoimmune diabetes in adults(LADA)[26.08%(60/230)],and the fasting blood glucose(FPG)[(10.93±6.98)mmol/L]and glycosylated hemoglobin(HbA_(1c))[(10.63±3.01)%]were high.The average completion rate of the overall data of the first diagnosed patients in SZT1D was only 60%[(62.9±31.5)%]:the number of patients with overall data completion≥80%in SZT1D was only 50.2%(350/697);the number of patients with overall data completion≥80%in Shenzhen was less than that outside Shenzhen[44.3%(222/501)vs 65.3%(128/196),P<0.001].The key indicators with better completion rate of first-visit were disease course[76.2%(531/697)],age of onset[75.8%(528/697)],family history of diabetes[74.9%(522/697)],etc.,but none of them had a completion rate of more than 80%,and the diabetes self-management behavior assessment questionnaire and scale score were completely missing;the frequency of daily blood glucose monitoring[46.1%(231/501)vs 64.3%(126/196),P<0.001],current insulin regimen[44.3%(222/501)vs 63.3%(124/196),P<0.001],number of diabetic ketoacidosis(DKA)since the onset of the disease[45.7%(229/501)vs 64.8%(127/196),P<0.001]and the number of symptomatic hypoglycemia in the past 1 month[39.3%(197/501)vs 63.8%(125/196),P<0.001]were higher in Shenzhen than those reported outside Shenzhen.In addition,the probability of abnormal FPG and postprandial glucose(PPG)[5.2%(24/466);3.8%(19/236)]were low.The revisit rate was not high[17.2%(120/697)],and the revisit rate in Shenzhen was higher than that outside Shenzhen[22.6%(113/501)vs 3.6%(7/196),P<0.001].The first revisit rate was 16.2%(113/697)and the second revisit rate was seriously insufficient[1.0%(7/697)].Conclusions The data quality of T1DM patients recorded by SZT1D needs to be further improved.Improving the information interconnection between China-T1D and SZT1D,employing quality control personnel and building a systematic data quality evaluation analysis and feedback mechanism are methods to promote the comprehensive,accurate and efficient input of T1DM data and continuously improve the evaluation methods to improve the overall data quality.