Study On The Cluster Classification Of Inpatients With Type 2 Diabetes And Its Relationship With Related Complications

Objective:This study takes inpatient T2 DM patients as the research object,uses clinically simple and easy-to-measure indicators as clustering variables for data-driven cluster analysis,and divides inpatient T2 DM patients into new clustering subtypes,aiming to verify the recentness of the Swedish Lief Groop team The stability of the proposed diabetes clustering in hospitalized T2 DM patients and the difference in the prevalence of diabetes-related complications among subtypes provide a theoretical basis for early secondary prevention of diabetes.Methods:2650 patients with T2 DM hospitalized in the First Hospital of Shanxi Medical University from January 2012 to December 2017 were included as the research object,which were divided into newly diagnosed T2 DM patients(course duration ≤1 year,n =1057),non-newly diagnosed T2 DM patients,Among them,patients with non-newly diagnosed T2 DM include patients with a short course(1 year <course ≤ 10 years,n = 957)and patients with a long course(10 years <course ≤ 20 years,n = 636).Collect patient general information,age at diagnosis of diabetes,body mass index(BMI),glycated hemoglobin(HbA1c),fasting blood glucose(FPG),fasting insulin(FINS),fasting C peptide(FCP),systolic blood pressure(SBP),diastolic blood pressure(DBP)),Blood lipid profile(TC,TG,HDL-c,LDL-c),cystatin c(Cys-c)and other related indicators and diabetic peripheral neuropathy(DPN),diabetic nephropathy(DN),diabetic retinopathy(DR),Atherosclerotic cardiovascular disease(ASCVD),non-alcoholic fatty liver(NAFLD),hypertension(HTN),hyperlipidemia and other related complications of diagnosis information.The diagnosis age of patients,BMI,HbA1 c,homogeneous model were used to evaluate the insulin resistance index(HOMA2-IR)and insulin secretion index(HOMA2-B)as clustering variables for k-means clustering analysis,and T2 DM patients were divided into new The clustering subtypes respectively verify the stability of the new clustering classification in the newly diagnosed and non-newly diagnosed patients,andcompare the differences in the prevalence of diabetes-related complications among the subtypes.Results:1.Analysis results of newly diagnosed inpatients with T2DM:(1)Diabetes clustering classification results:Newly diagnosed inpatients with type 2diabetes(n = 1057)were divided into 4 subtypes by k-means clustering analysis: mild obesity-related(MORD),severe insulin resistance(SIRD),and mild age-related(MARD),severe insulin deficiency(SIDD),the proportion of each subtype is 21.95%,25.17%,34.53%,18.35%.(2)Comparison of blood pressure,blood lipid profile,Cys-c and other indexes among cluster subtypes:SBP was the lowest in SIDD group(P <0.05);TG was the highest in MORD group(P <0.05);LDL and TC were not significant in each subgroup Difference(P> 0.05);Cys-c: SIRD group> MORD group> SIDD group(P <0.05).2.Analysis results of non-newly diagnosed T2 DM patients:2.1 Analysis results of short-term hospitalized T2 DM patients:(1)Diabetes clustering classification results: short-course inpatients patients with type2 diabetes(n = 957)were divided into three subtypes by k-means clustering: severe insulin resistance(SIRD),mild age-related(MARD)),Severe insulin deficiency type(SIDD),each subtype accounted for 32.92%,35.84%,%,31.24%.(2)Comparison of the prevalence of diabetes-related complications among various subtypes:Among the three subtypes of short-term hospitalized T2 DM patients,the prevalence of diabetic peripheral neuropathy(DPN)is the highest in SIDD,but there is no significant difference between the groups;the prevalence of diabetic nephropathy(DN)is in the SIDD group(66/299,22.07%)was significantly higher than the MARD group(45/343,13.12%)(P <0.05);the prevalence of diabetic retinopathy(DR)was significantly higher in the SIDD group(64/299,21.4%)than the MARD group(41 / 343,11.95%)(P<0.05);the prevalence of NAFLD was highest in the SIRD group(232 / 315,73.65%)(P<0.05);the prevalence of hypertension(HTN)was in MARD(168 / 343,48.98%),SIRD(155/315,49.21%)group was significantly higher than SIDD(103/299,34.45%)(P <0.05);hyperlipidemia was highest in the SIRD group(169/315,53.65%)(P < 0.05);The prevalence of cardiovascular disease(CVD)in the MARD group(76/343,22.16%)was significantly higher than that in the SIDD group(42/299,14.04%)(P <0.05).2.2.Analysis results of long-term inpatients T2 DM patients:(1)Diabetes clustering classification results: long-term inpatients T2 DM patients(n= 636)were divided into three subtypes by k-means clustering: severe insulin resistance(SIRD),mild age-related(MARD)),Severe insulin deficiency type(SIDD),each subtype accounted for 28.46%,41.19%,30.35%.(2)Comparison of the prevalence of diabetes-related complications among various subtypes:Among the 3 subtypes of long-term hospitalized T2 DM patients,the prevalence of DPN was slightly higher in SIDD,the prevalence of CVD was slightly higher in MARD,and the prevalence of hyperlipidemia was slightly higher in SIRD,but the difference was not statistically significant.The prevalence of other complications is not the same,the difference is statistically significant.The prevalence of DN in SIRD group(60/181,33.15%)was significantly higher than that in MARD group(58/262,22.14%)(P <0.05),the prevalence of DR in SIDD group(96/193,49.74%),The SIRD group(82/181,45.30%)was significantly higher than the MARD group(60/262,22.90%)(P <0.05);the prevalence of NAFLD was highest in the SIRD group(104/181,57.46%)(P <0.05);The prevalence of hypertension in the MARD(154/262,58.78%)and SIRD(112/181,61.88%)groups was significantly higher than that of SIDD(91/193,47.15%)(P <0.05).Conclusion:The newly proposed diabetes subtype of the Lief Groop team in Sweden has good stability and applicability in newly diagnosed inpatients with type 2 diabetes.Although there are migrations among clustered subtypes in inpatient T2 DM patients with non-new diagnosis(including short course and long course),the difference in the prevalence of diabetes-related complications among different subtypes is consistent with the results of the Lief Groop team.This new cluster classification can provide a theoretical basis for the early targeted prevention and treatment of diabetes-related complications,which will be beneficial to the early secondary prevention of diabetes.