Improved survival prediction from lung function data in a large population sample
Improved survival prediction from lung function data in a large population sample
复制标题
DOI：
10.1016/j.rmed.2008.09.016
复制
发表时间：
2009-03-01
影响因子：
4.3
通讯作者：
Vestbo, Jorgen
中科院分区：
文献类型：
--
作者：
Miller, Martin R.;Pedersen, Ole F.;Vestbo, Jorgen
有关肺功能与生存率的研究通常将肺功能损害表示为预测值的百分比，但这保留了年龄，身高和性别偏见。我们研究了哥本哈根城市心脏研究数据中预测全因和气道相关肺病死亡率的1秒用力呼气量（FEV 1）的替代方法。考克斯回归模型推导了13,900例受试者25年以上的生存率。入组时年龄、性别、吸烟状况、体重指数、既往心肌梗死和糖尿病是假定的预测因素，FEV 1作为原始数据，以身高幂（FEV 1/ht（n））标准化，作为预测百分比（FEV 1 PP）或作为标准化残差（FEV 1 SR）。在多变量模型中，FEV 1/ht（2）的五分位数在预测全因死亡率方面优于FEV 1 PP和FEV 1 SR，最差五分位数的风险比（HR）分别为2.8、2.0和2.1。使用肺功能的切割水平对损伤进行分类，最差类别FEV 1/ht（2）的全因和气道相关肺病死亡率的多变量预测HR分别为10和2044，而最差类别FEV 1 PP的HR分别为5和194。在全因死亡率的单变量预测中，FEV 1/ht（2）类别的HR比FEV 1 PP高2-4倍，气道相关性肺病死亡率高3-10倍。FEV 1/ht（2）比FEV 1 PP上级预测患者的生存率。在一般人群中，这种表达FEV 1损伤的方法最能反映随后死亡的风险。(C)2008爱思唯尔有限公司保留所有权利。
Studies relating tung function to survival commonly express lung function impairment as a percent of predicted but this retains age, height and sex bias. We have studied alternative methods of expressing forced expiratory volume in 1 s (FEV1) for predicting all cause and airway related lung disease mortality in the Copenhagen City Heart Study data. Cox regression models were derived for survival over 25 years in 13,900 subjects. Age on entry, sex, smoking status, body mass index, previous myocardial infarction and diabetes were putative predictors together with FEV1 either as raw data, standardised by powers of height (FEV1/ht(n)), as percent of predicted (FEV1PP) or as standardised residuals (FEV1SR). Quintiles of FEV1/ht(2) were better at predicting all, cause mortality in multivariate models than FEV1PP and FEV1SR, with the hazard ratio (HR) for the worst quintiles being 2.8, 2.0 and 2.1 respectively. Cut levels of lung function were used to categorise impairment and the HR for multivariate prediction of all cause and airway related lung disease mortality were 10 and 2044 respectively for the worst category of FEV1/ht(2) compared to 5 and 194 respectively for the worst category of FEV1PP. In univariate predictions of all cause mortality the HR for FEV1/ht(2) categories was 2-4 times higher than those for FEV1PP and 3-10 times higher for airway related tung disease mortality. We conclude that FEV1/ht(2) is superior to FEV1PP for predicting survival. in a general population and this method of expressing FEV1 impairment best reflects hazard for subsequent death. (C) 2008 Elsevier Ltd. All rights reserved.