背景：糖尿病是以胰岛素分泌缺陷和/或胰岛素作用缺陷为主要病理改变，以高血糖为主要特征并伴有多种异常的临床代谢综合征[1]。糖尿病是一个异质性疾病。既往的分类方法把糖尿病的临床类型分为：1型糖尿病、2型糖尿病、特殊类型糖尿病及妊娠糖尿病[1]，而其中占比最多的2型糖尿病具有高度的异质性，因此既往的分型方法经常给临床诊断及治疗带来不便。目前，国际上最新提出的对初诊糖尿病患者进行以6个变量为基础的聚类分析可以从另外一个角度对成人发病的糖尿病进行亚型分类，进而通过识别不同亚型间发生糖尿病并发症的风险的高低以提高精准治疗的可能，但这种分型方式是否适合在国人应用及对住院成人糖尿病患者的适用性有待研究。
目的：探讨成人发病的糖尿病的Emma Ahlqvist教授的国际新分型方法在住院糖尿病患者中的应用，同时探讨一下按照这种分型的方法不同亚型间糖尿病并发症的发生风险是否不同。
方法：选取2017年1月-2018年12月年间在北京医院内分泌科住院的成人糖尿病患者符合研究入组标准者共1152例，以住院期间谷氨酸脱羧酶抗体（GADA,glutamic acid decarboxylase antibodies）、体重指数（BMI）、糖化血红蛋白（HbA1c）、稳态模型评估的β细胞功能（HOMA-β%）及胰岛素抵抗指数（HOMA-IR）和糖尿病诊断年龄为变量，进行k均值聚类分析，确定成人发病的糖尿病的国际新分型亚型。使用卡方检验的方法判断新型亚型与糖尿病并发症：DN（diabetic nephropathy，糖尿病肾病）、DR（diabetic retinopathy，糖尿病视网膜病变）、DPVD(diabetic peripheral vascular disease，糖尿病周围血管病变)、DSPN(diabetic distal symmetric polyneuropathy，糖尿病远端对称性多神经病变)、DK（diabetic ketosis，糖尿病酮症）或伴发疾病：高血压（hypertension）、血脂异常（dyslipidemia）、CHD(coronary heart disease，冠心病)间的关系。
结果：住院成人糖尿病患者在使用聚类分析后可获得与国际相一致的5类亚型，其中轻度肥胖相关型糖尿病（MOD）患者人数最多占所有患者的34.55%,其它依次是轻度年龄相关型糖尿病（MARD）（21.55%）、严重胰岛素缺乏型糖尿病（SIDD）（20.51%）、严重胰岛素抵抗型糖尿病（SIRD）（19.02%）、严重自身免疫性糖尿病（SAID）（4.36%）。
其中第严重自身免疫性糖尿病及严重胰岛素缺乏型糖尿病发生糖尿病酮症的风险最高。第严重胰岛素缺乏型糖尿病视网膜病变的风险最高。轻度年龄相关型发生糖尿病周围血管病的风险最高。

Background:Diabetes mellitus is a metabolic disorder characterized by the presence of hyperglycemia due to defective insulin secretion, defective insulin action or both. According to existing classification system, the common types of diabetes mellitus are type 1 diabetes, type2 diabetes, other specific types of diabetes and gestational diabetes[1]. But type 2 diabetes is highly heterogeneous which increases difficulties in differing and treating in clinical practice. The novel international classification method based on six variables using a cluster method provide a new view of differing adult-onset diabetes. In the meantime, the different risk of diabetic complications between subgroups could enhance the possibility of precise treatment. But the result among inpatient is uncertain.
Purpose: Discus whether the novel subgroups of adult-onset diabetes discovered by professor Emma Ahlqvist in hospitalized diabetes remains the same with the newly diagnosed diabetes. Discuss the different risk of developing a diabetic complication among different subgroups.
Methods: we collected inpatients data in Beijing Hospital from 2017, January to 2018, December who has diagnosed with type 1 diabetes, type 2 diabetes or LADA. Then, we use six variables:glutamate decarboxylase antibodies（GADA）, BMI, HbA1c, and homoeostatic model assessment  estimates of β-cell function and insulin resistance, age at diagnosis) to do a data driven analysis using k-means clustering method. And compare the risks for each subgroup developing a diabetic complication,including DN（diabetic nephropathy）, DR(diabetic retinopathy),DPVD(diabetic peripheral vascular disease), DSPN(diabetic  distal symmetric polyneuropathy), DK（diabetic ketosis）or a concomitant disease such as hypertension, dyslipidemia, CHD(coronary heart disease） with a chi-square test.
Findings:After clustering there are five subgroups exactly the same as the previous findings: the biggest part is mild obesity-related diabetes(34.55%), follwed as mild age-related diabetes(21.55%), severe insulin-deficient diabetes(20.51%),severe insulin-resistant diabetes(19.02%) and severe autoimmune diabetes(4.36%).
In particular,severe autoimmune diabetes and severe insulin-suficient diabetes have significant higher risk of diabetic ketosis. Severe insulin-suficient diabetes has significant higher risk of diabetic retinopathy. Mild age-related diabetes has significant higher risk of diabetic peripheral vascular disease.