摘要目的:分析深圳市≥40岁社区居民高同型半胱氨酸（Hcy）的患病情况及其与脑卒中患病风险的相关性。方法:2012年3月—2015年7月，采用整群随机抽样的方法选取深圳市3个社区筛查点16 792例脑卒中筛查者作为研究对象。运用Lasso方法对影响脑卒中发生的相关变量进行选择，利用交叉验证选取模型中最优调和参数λ，再建立Logistic回归模型分析高Hcy与脑卒中患病的关联性，并根据年龄、性别、吸烟状况、体育锻炼情况、高血压史、卒中家族史等因素进行分层分析及相乘交互作用检验。结果:在16 792例筛查者中，合计有2 992例高Hcy患者和984例脑卒中患者，两者患病率分别为17.82%和5.86%。交叉验证选择的最优λ为0.009，经Lasso回归模型筛选出影响脑卒中发生的主要变量有7个。拟合Logistic回归模型显示：高Hcy与脑卒中患病显著相关（ OR=1.395，95% CI：1.196~1.627， P<0.001），随着Hcy水平升高，脑卒中患病风险呈增高趋势。相乘交互作用结果显示：Hcy与年龄、高血压史等因素在影响脑卒中患病上存在显著的交互效应（均 P<0.05），进一步分层分析显示，在年龄<55岁、具有高血压史的人群中，高Hcy与脑卒中的患病风险更高。 结论:深圳市≥40岁社区居民高Hcy的患病率相对较低，高Hcy与脑卒中患病密切相关，且年龄和高血压史等因素能影响高Hcy与脑卒中的患病风险。

abstractsObjective:To analyze the prevalence of high homocysteine (Hcy) in community residents aged 40 years and above in Shenzhen and its correlation with the risk of stroke.Methods:Between March 2012 and July 2015, a cluster sampling method was used to select 16 792 screened population from three community screening points in Shenzhen city to serve as the research subjects. Lasso method was used to select the variables related to the incidence of stroke. Cross validation method was used to choose the optimal lambda in the model. Logistic regression model was established to analyze the association between high Hcy and stroke. Interaction and stratified analyses were conducted according to age, sex, smoking status, physical activity, history of hypertension and family history of stroke.Results:Among the 16 792 screened population, 2 992 patients with high Hcy and 984 stroke patients were found, and the prevalence was 17.82% and 5.86%, respectively. The optimal lambda selected by cross validation was 0.009, and 7 main variables influencing the incidence of stroke were identified by Lasso regression model. Multiple logistic regression analysis showed that high Hcy ( OR=1.395, 95% CI: 1.196—1.627, P<0.001) was significantly associated with stroke. With the increase of Hcy levels, the risk of stroke increased gradually. The interaction analysis revealed that age and history of hypertension played an interactive role in the association between high Hcy and stroke. Further stratified analysis found that the risk of high Hcy and stroke were higher in the participants who were younger than 55 years old and with a history of hypertension. Conclusions:The prevalence of high Hcy in community residents aged 40 years and above in Shenzhen is relatively low. High Hcy is closely correlated with stroke, and the factors such as age and history of hypertension can affect the risk of high Hcy and stroke.