基于纵向数据全年龄段肺通气正常值统计模型构建及应用研究

Reference value is an important valuation index of lung function test in clinical diagnosis and disease prevention. The clear understanding of change of lung function of healthy people due to time change is essential for correctly interpreting the change of lung function due to disease or treatment. However, cross-sectional design were used in most previous studies, which were lack of repeating data of short-term or long-term spirometric data; And they were established based on only children or adults population separately, which usually introduce bias and even set up by incorrect statistical analysis model. Currently there are lack of effective statistical methods to establish spirometric reference values across all ages with longitudinal data. Therefore, this study proposes a new semi-parametric regression model, and the parameters component and nonparametric component were estimated by improved empirical likehood and maximum penalized likelihood separately. Asymptotic properties of the resulting estimators such as consistency and asymptotic normality were established. This project will then validate the feasibility and effectiveness of the methods by Monte Carlo simulation studies, and finally validate the model by healthy spirometric data from cohort study. This study will help prevention and treatment of respiratory diseases, especially in the early diagnosis of important guiding significance, and provide statistical methods for similar data to study reference.

正常参考值是肺功能检查临床诊断和疾病防治重要的参考指标，要想正确地解释由于疾病或治疗引起的肺功能改变，需清楚了解健康人肺功能随时间变化而出现的相应改变。然而以往研究大多数采用横断面设计，缺乏短期或长期的肺通气重复测试数据；而且分别基于儿童或者成人人群单独建立，容易引入偏倚；甚至采用错误统计分析方法来建模。目前缺乏有效的统计方法来建立基于纵向数据全年龄段肺通气正常参考值。因此，本研究提出一种新的半参数回归模型，参数分量和非参数分量估计量分别拟用改良经验似然法和最大惩罚似然法来估计，并推导证明估计具有相合性和渐近正态性，然后通过模拟研究和临床健康人肺通气随访调查数据来验证该模型的可行性和有效性。本研究结果可为肺通气正常参考值的建立提供一种有效统计方法，使得临床上可更有针对性地采取干预和预防，对呼吸系统疾病的防治，尤其是早期诊断有非常重要的指导意义，也为同类数据提供统计方法学参考。

肺功能正常参考值的选择和建立,是肺功能检查临床诊断和疾病防治的基本依据。要想正确地解释由于疾病或治疗引起的肺功能改变，需清楚了解健康人肺功能随时间变化而出现的相应改变，然而目前缺乏一种有效的统计方法来建立基于纵向数据全年龄段肺通气正常参考值。本研究在GAMLSS模型基础上，构建了一种基于纵向数据的全年龄段肺通气正常参考值的半参数回归模型，推导证明估计具有相合性和渐近正态性，并开发相应的可视化程序；模拟研究结果显示：和其他估计方法相比，改良后的GAMLSS模型估计结果较好，有更小的偏差和根均方误差，且随着样本量增加而减小；最后利用随访研究收集健康人肺通气数据、全国第二次肺功能调查研究数据和临床实践指南制订后不同年份随访的质量评价相关的纵向数据来验证该模型的可行性和有效性。本研究还提出一个适合于纵向数据的肺通气功能正常参考值建立的样本量估算公式，模拟研究结果和实际应用结果显示所估算出的样本量能够达到预期功效。该研究结果为肺通气正常参考值的建立提供一种有效统计方法，使得临床上可更有针对性地采取干预和预防，对呼吸系统疾病的防治，尤其是早期诊断有非常重要的指导意义，也为其他纵向数据提供一种有效的分析方法。