中文摘要：

目的探讨潜在类别模型在超声诊断的常见慢性病与人群日常生活行为方式分型关联性研究中的应用。方法采用潜在类别分析方法分析早餐是否规律、吸烟、被动吸烟、大量饮酒、锻炼情况等生活行为模式,根据人群生活行为方式对研究人群聚类后,比较不同行为方式人群的人口学特征及B超诊断常见慢性病分布情况。结果 1164例研究对象的行为方式可分为3大类,分别是行为方式基本健康组、行为方式亚健康组和行为方式不健康组。各组人数分布情况依次为：208人（17.3%）、223人（21.4%）和733人（61.3%）。三组人群的年龄、性别、血型、婚姻状况、文化程度和职业状况的内部构成不同（所有P〈0.05）。本研究结果显示：除脂肪肝在3类行为方式组中的分布不同外,其他B超诊断常见慢性病的分布无统计学差异（P〉0.05）。大多数诊断结果（除肾结石、脂肪肝、肝囊肿外）的变化趋势均符合从行为方式基本健康组、行为方式亚健康组、行为方式不健康组依次增高的变化趋势。多因素logistic回归分析结果表明,影响脂肪肝的主要因素有肥胖、吸烟、大量饮酒及低豆制品摄入。结论潜在类别模型可用于人群的日常行为方式分型研究,探索不同行为方式与慢性病之间的关联性,揭示健康行为干预的重点人群和内容,进而为有针对性制定干预措施提供科学依据。

英文摘要：

Objective To investigate the application of latent class model in the research on the correlation between common chronic diseases diagnosed by ultrasound and lifestyle. Methods Latent class analysis（ LCA） was used to indentify patterns of health behavior among eating breakfast,smoking,passive smoking,heavy drinking and exercise,and to classify the individuals. Further studies were conducted to compare demographic characteristics and the distributions of common chronic diseases diagnosed by ultrasound in different subgroups. Results Out of 1164 subjects,three subgroups of health behavior were identified which included ＂healthy behavioral patterns＂（ 17. 4%）,＂sub-healthy behavioral patterns＂（ 21. 3%） and ＂unhealthy behavioral patterns＂（ 61. 3%）. The internal constituents of sex,blood type,marital status,educational level and occupational status were different（ all P〈0. 05）. No differences of the distribution of common chronic diseases diagnosed by ultrasound were found except for fatty liver among three behavior subgroups. Cholecystitis and gallbladder removal have a larger proportion in the subhealth behavior groups and（ or） unhealthy behavior group than that in the healthy group diagnosed by type-B sonic. Except for kidney stones,fatty liver and liver cyst,most chronic diseases have the largest distribution in the unhealthy subgroup. Multi-logistic regression Results show ed that risk factors associated with fatty liver were Obesity,smoking,heavy drinking and low soy food intake. Conclusions Latent class model can be applied in the classification of health behavior patterns,which can indicate the key health behavior intervention and provide scientific evidence for the intervention measures.